feat: agent runtime online + minimal safety spine (runtime root)#2003
Conversation
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.Scanned FilesNone |
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request completes the critical-path work for the agent runtime root by introducing a conditional initialization mechanism. It ensures that the agent engine is only constructed when a valid provider is present, providing a robust 'empty-company' mode that prevents task submission failures and improves observability. The changes include a new deterministic test driver for simulation, enhanced concurrency controls for service hot-swapping, and improved workspace persistence configuration. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI (base), Organization UI (inherited) Review profile: ASSERTIVE Plan: Pro Run ID: 📒 Files selected for processing (4)
📜 Recent review details⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (7)
🧰 Additional context used📓 Path-based instructions (5)tests/**/*.py📄 CodeRabbit inference engine (CLAUDE.md)
Files:
⚙️ CodeRabbit configuration file
Files:
!(providers/presets.py|web/public/provider-logos/**|.claude/**)📄 CodeRabbit inference engine (CLAUDE.md)
Files:
src/synthorg/**/*.py📄 CodeRabbit inference engine (CLAUDE.md)
Files:
src/synthorg/{workers,api}/**/*.py📄 CodeRabbit inference engine (CLAUDE.md)
Files:
src/**/*.py⚙️ CodeRabbit configuration file
Files:
🧠 Learnings (2)📓 Common learnings📚 Learning: 2026-05-05T09:04:46.195ZApplied to files:
🔇 Additional comments (4)
WalkthroughThis PR selects and installs a worker execution service at startup (provider-present vs empty-company), adds AppState workspace-root pinning and a hot-swap seam, introduces a deterministic scripted CompletionProvider, implements AgentEngineExecutionService and NoProviderExecutionService, blocks task creation when no provider exists (409 / AGENT_RUNTIME_NOT_CONFIGURED), and adds runtime builder wiring plus E2E/integration/unit tests validating install ordering, empty-company rejection, and service swap behavior. |
Aureliolo
temporarily deployed
to
cloudflare-preview
GitHub Actions
Inactive
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 5
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/synthorg/api/app_helpers.py`:
- Around line 144-147: The code in app_helpers.py reads SYNTHORG_DB_PATH into
db_path and returns Path(db_path).parent / "agent-workspaces" without checking
if db_path is absolute; update the logic in the function that contains db_path
so you validate that db_path is an absolute path (use os.path.isabs or
Path(db_path).is_absolute()) before deriving the workspace root, and only return
Path(db_path).parent / "agent-workspaces" when it is absolute; if it is missing
or not absolute, fall back to the existing SYNTHORG_DATABASE_URL branch or
another safe absolute default.
In `@tests/e2e/test_runtime_online_seam.py`:
- Around line 94-95: Update the test function signatures to use a concrete Path
type for the tmp_path parameter instead of object (e.g., change "tmp_path:
object" to "tmp_path: Path") and remove any "# type: ignore[arg-type]" comments;
ensure you import Path from pathlib at the top of the file if not already
present and apply the same change for the second occurrence around the other
test (the tmp_path parameter at lines ~140-141) so type checking is preserved
for the tmp_path seam.
In `@tests/integration/api/test_task_create_empty_company.py`:
- Around line 27-29: The header-name check in the loop over
resp.headers.multi_items() is case-sensitive and can miss Set-Cookie variants;
update the condition that currently tests k != "set-cookie" to perform a
case-insensitive comparison (e.g., compare k.lower() or k.casefold() to
"set-cookie") so that any capitalization of the Set-Cookie header is accepted
while still continuing for non-cookie headers; keep the loop and variables k and
v unchanged except for the updated comparison.
In `@tests/unit/api/test_state_swap_worker_execution_service.py`:
- Around line 56-57: The test is currently catching any Exception for the access
to state.worker_execution_service; change the assertion to expect the specific
domain/service exception that the worker execution retrieval raises (replace
Exception with the actual exception class thrown, e.g., TaskEngineError or
ServiceUnavailableError) so the test only passes for the intended failure path;
update the pytest.raises(...) wrapper around state.worker_execution_service to
reference that specific exception type (locate the exception class in the task
engine / service module and import it into the test).
In `@tests/unit/providers/drivers/test_scripted.py`:
- Around line 128-131: The test currently calls structlog.reset_defaults()
directly, which mutates global state and can leak into other tests; change the
test to restore the previous structlog configuration after the assertion (or
call reset inside a finally block) so global structlog settings are not left
modified—wrap the reset/use of
ScriptedDriver(strategy=DeterministicResponseStrategy()) and capture_logs() in a
try/finally that restores the original structlog configuration (or move reset
into a teardown fixture) to ensure no process-level state leakage across tests.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI (base), Organization UI (inherited)
Review profile: ASSERTIVE
Plan: Pro
Run ID: b2dc969c-1524-48b2-92ad-88e50f3dedce
📒 Files selected for processing (33)
CLAUDE.mdscripts/_ghost_wiring_manifest.txtsrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/api/state.pysrc/synthorg/api/state_services.pysrc/synthorg/core/domain_errors.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/provider.pysrc/synthorg/observability/events/workers.pysrc/synthorg/providers/drivers/scripted.pysrc/synthorg/providers/registry.pysrc/synthorg/workers/execution_service.pysrc/synthorg/workers/runtime_builder.pytests/_shared/scripted_provider.pytests/e2e/conftest.pytests/e2e/test_runtime_online_seam.pytests/integration/api/conftest.pytests/integration/api/test_runtime_install_ordering.pytests/integration/api/test_task_create_empty_company.pytests/integration/engine/test_coordination_wiring.pytests/unit/api/conftest.pytests/unit/api/controllers/test_coordination.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/engine/quality/scripted_provider.pytests/unit/providers/drivers/test_scripted.pytests/unit/workers/test_execution_service.pytests/unit/workers/test_runtime_builder.pyweb/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
📜 Review details
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (6)
- GitHub Check: Dashboard Test
- GitHub Check: Test Unit
- GitHub Check: Test Integration
- GitHub Check: Build Web Assets (melange)
- GitHub Check: Lighthouse Dashboard
- GitHub Check: Analyze (python)
🧰 Additional context used
📓 Path-based instructions (13)
src/synthorg/!(persistence)/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Only
src/synthorg/persistence/may import sqlite/psycopg or emit raw SQL
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/providers/registry.pysrc/synthorg/observability/events/api.pysrc/synthorg/api/app_helpers.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/domain_errors.pysrc/synthorg/api/app.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/state_services.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/workers/execution_service.pysrc/synthorg/providers/drivers/scripted.py
src/synthorg/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
No hardcoded numeric values; numerics live in
settings/definitions/; allowlist only 0/1/-1, HTTP codes, hex masks, powers-of-2, and module-level annotated named constantsComments must explain WHY only; no reviewer citations / issue back-refs / migration framing; enforced by
check_no_review_origin_in_code.pyandcheck_no_migration_framing.pyNo
from __future__ import annotations(Python 3.14 has PEP 649)Type hints required on public functions; mypy strict; Google-style docstrings; line length 88; functions <50 lines; files <800 lines
Errors must use
<Domain><Condition>Errorpattern inheriting fromDomainError, never fromException/RuntimeErrordirectly; enforced bycheck_domain_error_hierarchy.pyPydantic v2 models must be frozen with
extra='forbid'on every frozen model project-wide;@computed_fieldis auto-exempt; use per-line# lint-allow: frozen-extra-forbid -- <reason>for boundaries; enforced bycheck_frozen_model_extra_forbid.pyUse
@computed_fieldfor derived fields in Pydantic modelsUse
NotBlankStrfor identifiers in Pydantic modelsRequire args models at every system boundary; use
parse_typed()for every external dict ingestion; enforced bycheck_boundary_typed.pyFor immutability, use
model_copy(update=...)orcopy.deepcopy(); deepcopy at system boundariesUse
asyncio.TaskGroupfor fan-out/fan-in; async helpers must catchExceptionand re-raiseMemoryError/RecursionErrorUse Clock seam pattern:
clock: Clock | None = None; tests injectFakeClockLifecycle services must own
_lifecycle_lock; timed-out stops mark unrestartableUntrusted content (SEC-1) must use
wrap_untrusted()fromengine.prompt_safety; useHTMLParseGuardfor HTMLImport logger via
from synthorg.observability import get_logger; variable must always belogger; never useimport loggingorprint()in app codeEvent names must come from
observability.events.<domain>constants; use structured kwargs (`logger.info...
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/providers/registry.pysrc/synthorg/observability/events/api.pysrc/synthorg/api/app_helpers.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/domain_errors.pysrc/synthorg/api/app.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/state_services.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/workers/execution_service.pysrc/synthorg/providers/drivers/scripted.py
src/**/*.py
⚙️ CodeRabbit configuration file
This project uses Python 3.14+ with PEP 758 except syntax: "except A, B:" (comma-separated, no parentheses) is correct and mandatory -- do NOT flag it as a typo or suggest parenthesized form. The "except builtins.MemoryError, RecursionError: raise" pattern is intentional project convention for system-error propagation. When evaluating the 50-line function limit, count only the function body excluding the signature lines, decorators, and docstring. Functions 1-5 lines over due to docstrings or multi-line signatures should not be flagged. Do not suggest extracting single-use helper functions called exactly once -- this reduces readability without improving maintainability.
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/providers/registry.pysrc/synthorg/observability/events/api.pysrc/synthorg/api/app_helpers.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/domain_errors.pysrc/synthorg/api/app.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/state_services.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/workers/execution_service.pysrc/synthorg/providers/drivers/scripted.py
src/synthorg/providers/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Provider calls must go through
BaseCompletionProvider(retry + rate limit); never implement retry in driver subclasses; retryable:RateLimitError,Provider{Timeout,Connection,Internal}Error
Files:
src/synthorg/providers/registry.pysrc/synthorg/providers/drivers/scripted.py
tests/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Use pytest markers:
@pytest.mark.{unit,integration,e2e,slow}; asyncauto; timeout 30s global; coverage 80% minWindows: unit tests must use
WindowsSelectorEventLoopPolicy(3.14 IOCP teardown race); subprocess tests override backTest doubles ladder:
FakeClockfor Clock seam,mock_of[T](**overrides)for typed-boundary substitutions,SimpleNamespacefor attribute-bags; bareMagicMockat typed boundary blocked byscripts/check_mock_spec.pyImport
FakeClockandmock_offromtests._shared; inject viaclock=and helper's spec subscriptHypothesis: 10 deterministic CI examples; failures are real bugs (fix + add
@example(...))Never skip/xfail flaky tests; fix fundamentally; use
asyncio.Event().wait()instead ofsleep(large)
Files:
tests/unit/workers/test_execution_service.pytests/integration/engine/test_coordination_wiring.pytests/unit/api/controllers/test_coordination.pytests/integration/api/test_runtime_install_ordering.pytests/unit/providers/drivers/test_scripted.pytests/unit/engine/quality/scripted_provider.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/conftest.pytests/unit/api/test_state_swap_worker_execution_service.pytests/e2e/test_runtime_online_seam.pytests/unit/workers/test_runtime_builder.pytests/_shared/scripted_provider.pytests/integration/api/conftest.pytests/e2e/conftest.py
⚙️ CodeRabbit configuration file
Test files do not require Google-style docstrings on classes or functions -- ruff D rules are only enforced on src/. A bare
@settings() decorator with no arguments on Hypothesis property tests is a no-op and should not be suggested -- the HYPOTHESIS_PROFILE env var controls example counts via registered profiles, which@given() honors automatically.
Files:
tests/unit/workers/test_execution_service.pytests/integration/engine/test_coordination_wiring.pytests/unit/api/controllers/test_coordination.pytests/integration/api/test_runtime_install_ordering.pytests/unit/providers/drivers/test_scripted.pytests/unit/engine/quality/scripted_provider.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/conftest.pytests/unit/api/test_state_swap_worker_execution_service.pytests/e2e/test_runtime_online_seam.pytests/unit/workers/test_runtime_builder.pytests/_shared/scripted_provider.pytests/integration/api/conftest.pytests/e2e/conftest.py
src/synthorg/api/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
agent_registrymust be built BEFOREauto_wire_meetingsduring API construction phase
tunnel_provideris wired unconditionally during construction phase (not gated byintegrations.enabled)On-startup:
SettingsServiceauto-wire must precedeWorkflowExecutionObserverregistration;OntologyServicewires afterpersistence.connect()via_wire_ontology_service
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/app.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/state_services.pysrc/synthorg/api/state.pysrc/synthorg/api/controllers/tasks.py
**/*.{d2,md}
📄 CodeRabbit inference engine (CLAUDE.md)
Use D2 for architecture/nested container diagrams with theme 200 (Dark Mauve), pinned to v0.7.1 in CI; use Mermaid for flowcharts/sequence/pipelines; use Markdown tables for tabular data
Files:
CLAUDE.md
web/src/**/*.{js,jsx,ts,tsx,mts}
📄 CodeRabbit inference engine (web/CLAUDE.md)
web/src/**/*.{js,jsx,ts,tsx,mts}: Always usecreateLoggerfrom@/lib/logger; never bareconsole.warn/console.error/console.debugin application code. Variable name must always belog. Onlylogger.tsitself may use bare console methods. Uselog.debug()(DEV-only, stripped in production),log.warn(),log.error().
Pass dynamic/untrusted values as separate args to logger calls (not interpolated into the message string) so they go throughsanitizeArg
Attacker-controlled fields inside structured objects must be wrapped insanitizeForLog()before embedding in log calls
Error-code constants (MANDATORY): importErrorCodeandErrorCategoryfrom@/api/types/errors(re-exported from the generatedweb/src/api/types/error-codes.gen.ts). Discriminate onErrorCode.<NAME>, never on raw integer literals.
Use@eslint-react/web-api-no-leaked-fetchto detectfetch()in effects withoutAbortControllercleanup
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/api/types/**/*.gen.ts
📄 CodeRabbit inference engine (web/CLAUDE.md)
Generated DTO types (MANDATORY): NEVER hand-edit
web/src/api/types/*.gen.ts. Regenerate withuv run python scripts/generate_dto_types_ts.py. Import DTOs via the barrel (import type { AgentConfig } from '@/api/types').
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/**/*.{ts,tsx,mts}
📄 CodeRabbit inference engine (web/CLAUDE.md)
web/src/**/*.{ts,tsx,mts}: Use@typescript-eslint/no-floating-promisesto forbid unawaited promises so async work cannot survive the test that scheduled it and trip the active-handle gate
Use@typescript-eslint/no-misused-promises(withchecksVoidReturn: { attributes: false }) to forbid passing async functions where the callsite ignores the returned promise. React 19asyncevent handlers stay allowed via theattributes: falseexemption.
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/**/*.{tsx,ts}
📄 CodeRabbit inference engine (CLAUDE.md)
Reuse
web/src/components/ui/design tokens only for Web Dashboard Design System; seeweb/CLAUDE.mdfor details
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
src/synthorg/api/controllers/setup/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Setup completion:
post_setup_reinit()must serialize provider reload, agent bootstrap, worker-execution-service rebuild underCOMPLETE_LOCKto prevent race conditions on setup_complete flag
Files:
src/synthorg/api/controllers/setup/agent_helpers.py
src/synthorg/workers/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Worker execution service: use
synthorg.workers.runtime_builder.build_worker_execution_servicebehind provider-present switch; install viaAppState.worker_execution_serviceseam; boot install hook appended FIRST after persistence/SettingsService hooks
Files:
src/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
🧠 Learnings (4)
📓 Common learnings
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Read `docs/design/` page before implementing; deviations need approval per DESIGN_SPEC.md
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Present every plan for accept/deny before coding
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: No region/locale should be privileged; use metric units; use British English for all user-facing text
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Every convention PR must ship its enforcement gate
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: After issue completion: branch + commit + push (no auto-PR); use `/pre-pr-review` before creating PR; after PR created use `/aurelio-review-pr` for external feedback
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Commits must follow `<type>: <description>` format (feat/fix/refactor/docs/test/chore/perf/ci); commitizen-enforced
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Signed commits required on protected refs (GPG/SSH or GitHub App via `synthorg-repo-bot`)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Branches must follow `<type>/<slug>` format from main
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: Squash merge; PR body becomes squash commit; trailers (`Release-As`, `Closes `#N``) must be in PR body
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: After every squash merge, run `/post-merge-cleanup`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:06:25.508Z
Learning: CLI is Docker-only (init/start/stop/status); features go in dashboard + REST API
📚 Learning: 2026-05-05T09:04:46.195Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1760
File: scripts/_dual_backend_parity_lib.py:215-216
Timestamp: 2026-05-05T09:04:46.195Z
Learning: This repository targets Python 3.14+ and follows PEP 758. Therefore, reviewer tooling should NOT treat unparenthesized multi-exception `except` clauses written without an `as` clause (e.g., `except MemoryError, RecursionError:`) as syntax errors. Only flag `except`-clause problems when they are genuinely invalid for Python 3.14+.
Applied to files:
src/synthorg/observability/events/provider.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/providers/registry.pysrc/synthorg/observability/events/api.pytests/unit/workers/test_execution_service.pysrc/synthorg/api/app_helpers.pytests/integration/engine/test_coordination_wiring.pytests/unit/api/controllers/test_coordination.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/domain_errors.pytests/integration/api/test_runtime_install_ordering.pytests/unit/providers/drivers/test_scripted.pysrc/synthorg/api/app.pytests/unit/engine/quality/scripted_provider.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/conftest.pytests/unit/api/test_state_swap_worker_execution_service.pytests/e2e/test_runtime_online_seam.pysrc/synthorg/api/controllers/setup/agent_helpers.pytests/unit/workers/test_runtime_builder.pysrc/synthorg/api/state_services.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pytests/_shared/scripted_provider.pysrc/synthorg/api/controllers/tasks.pytests/integration/api/conftest.pysrc/synthorg/workers/execution_service.pysrc/synthorg/providers/drivers/scripted.pytests/e2e/conftest.py
📚 Learning: 2026-05-16T18:36:31.446Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1944
File: docs/reference/conventions.md:787-789
Timestamp: 2026-05-16T18:36:31.446Z
Learning: In Aureliolo/synthorg, do not require adding `<!--RS:...-->` “Doc Numeric Claims (MANDATORY)” numeric macros for Python version numbers mentioned in documentation prose (e.g., “Python 3.14”, “Python 3.15”). The `scripts/check_doc_numeric_macros.py` gate only applies to `README.md`, `docs/index.md`, `docs/roadmap/index.md`, `docs/architecture/decisions.md`, and `docs/reference/convention-gates.md`, and it only flags digits adjacent to specific stat nouns (tests/providers/agents/stars/releases), not language version mentions like “Python 3.14”.
Applied to files:
CLAUDE.md
📚 Learning: 2026-05-16T18:36:35.250Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1944
File: docs/getting_started.md:109-109
Timestamp: 2026-05-16T18:36:35.250Z
Learning: When reviewing Markdown in the synthorg repo, account for the CI gate `check_doc_numeric_macros.py`: it skips fenced code blocks entirely, and it only flags digits that are adjacent to these stat nouns: `tests`, `providers`, `agents`, `stars`, `releases`. Therefore, numeric examples such as CLI flag values (e.g., `--num-workers=4` in fenced bash blocks) and prose version numbers (e.g., `3.14`/`3.15`) are not expected to trigger this check; prioritize changes only when digits appear next to one of the listed nouns (e.g., “5 tests”, “10 stars”, etc.).
Applied to files:
CLAUDE.md
🔇 Additional comments (36)
CLAUDE.md (1)
82-83: LGTM!scripts/_ghost_wiring_manifest.txt (1)
26-26: LGTM!src/synthorg/core/domain_errors.py (1)
248-264: LGTM!src/synthorg/core/error_taxonomy.py (1)
109-109: LGTM!src/synthorg/observability/events/api.py (1)
64-64: LGTM!src/synthorg/observability/events/provider.py (1)
9-9: LGTM!src/synthorg/observability/events/workers.py (1)
72-81: LGTM!web/src/api/types/error-codes.gen.ts (1)
57-57: LGTM!web/src/api/types/openapi.gen.ts (1)
7295-7295: LGTM!src/synthorg/workers/execution_service.py (5)
1-63: LGTM!
66-93: LGTM!
96-192: LGTM!
195-348: LGTM!
351-386: LGTM!tests/unit/workers/test_execution_service.py (4)
1-31: LGTM!
34-44: LGTM!
47-130: LGTM!
132-237: LGTM!src/synthorg/providers/drivers/scripted.py (1)
1-293: LGTM!src/synthorg/providers/registry.py (1)
141-146: LGTM!tests/_shared/scripted_provider.py (1)
1-269: LGTM!tests/unit/engine/quality/scripted_provider.py (1)
1-19: LGTM!tests/e2e/conftest.py (1)
1-38: LGTM!tests/integration/api/conftest.py (1)
12-94: LGTM!tests/integration/api/test_runtime_install_ordering.py (1)
1-57: LGTM!tests/integration/api/test_task_create_empty_company.py (1)
37-112: LGTM!tests/integration/engine/test_coordination_wiring.py (1)
285-302: LGTM!tests/unit/api/conftest.py (1)
30-30: LGTM!Also applies to: 46-47, 356-370, 426-427, 459-460
tests/unit/api/controllers/test_coordination.py (1)
127-144: LGTM!src/synthorg/workers/runtime_builder.py (1)
43-144: LGTM!tests/unit/workers/test_runtime_builder.py (1)
26-150: LGTM!src/synthorg/api/state.py (1)
176-177: LGTM!Also applies to: 428-441, 858-935, 953-958
src/synthorg/api/state_services.py (1)
137-137: LGTM!Also applies to: 451-490, 527-531
src/synthorg/api/app.py (1)
33-34: LGTM!Also applies to: 980-1029
src/synthorg/api/controllers/tasks.py (1)
25-25: LGTM!Also applies to: 37-37, 182-188
src/synthorg/api/controllers/setup/agent_helpers.py (1)
142-166: LGTM!
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request implements the "provider-present switch" for the agent runtime, enabling the system to select between a live AgentEngineExecutionService and a NoProviderExecutionService backstop based on LLM provider availability. It introduces the AgentRuntimeNotConfiguredError (4014) to reject task creation in unconfigured environments and adds a deterministic ScriptedDriver for testing and simulation. The changes also support hot-swapping execution services during setup and environment-driven resolution of agent workspace roots. Review feedback focuses on adhering to the "no-hardcoded-values" style rule by defining constants for workspace paths, improving error specificity for unassigned tasks, and ensuring the AutonomyResolver correctly utilizes the application's security configuration.
| def resolve_agent_workspace_root_env() -> Path | None: | ||
| """Resolve the agent sandbox workspace root from the environment. | ||
|
|
||
| Returns ``<runtime data dir>/agent-workspaces`` when an env-driven | ||
| deployment is in effect, so the agent's file-system / sandbox tools | ||
| write onto the mounted data volume rather than a process temp dir. | ||
| Returns ``None`` for injected / dev apps (no deployment env vars), | ||
| where :attr:`AppState.agent_workspace_root` keeps its documented | ||
| process-stable temp fallback. | ||
|
|
||
| Precedence mirrors the persistence env resolution: | ||
| ``SYNTHORG_ARTIFACT_DIR`` (explicit), then ``SYNTHORG_DB_PATH`` | ||
| parent (sqlite volume), then ``/data`` when only | ||
| ``SYNTHORG_DATABASE_URL`` is set (postgres compose volume). | ||
| """ | ||
| artifact_dir = os.environ.get("SYNTHORG_ARTIFACT_DIR", "").strip() | ||
| if artifact_dir: | ||
| return Path(_resolve_artifact_dir_env()) / "agent-workspaces" | ||
| db_path = os.environ.get("SYNTHORG_DB_PATH", "").strip() | ||
| if db_path: | ||
| return Path(db_path).parent / "agent-workspaces" | ||
| if os.environ.get("SYNTHORG_DATABASE_URL", "").strip(): | ||
| return Path("/data") / "agent-workspaces" | ||
| return None |
There was a problem hiding this comment.
The string "agent-workspaces" is hardcoded and repeated three times in this function. Additionally, "/data" is a hardcoded path. Per the "no-hardcoded-values" rule in CLAUDE.md, these should be defined as constants to improve maintainability and adhere to project standards.
_AGENT_WORKSPACE_DIR: Final[str] = "agent-workspaces"
_DEFAULT_DATA_DIR: Final[str] = "/data"
def resolve_agent_workspace_root_env() -> Path | None:
"""Resolve the agent sandbox workspace root from the environment.
Returns <runtime data dir>/agent-workspaces when an env-driven
deployment is in effect, so the agent's file-system / sandbox tools
write onto the mounted data volume rather than a process temp dir.
Returns None for injected / dev apps (no deployment env vars),
where AppState.agent_workspace_root keeps its documented
process-stable temp fallback.
Precedence mirrors the persistence env resolution:
SYNTHORG_ARTIFACT_DIR (explicit), then SYNTHORG_DB_PATH
parent (sqlite volume), then /data when only
SYNTHORG_DATABASE_URL is set (postgres compose volume).
"""
artifact_dir = os.environ.get("SYNTHORG_ARTIFACT_DIR", "").strip()
if artifact_dir:
return Path(_resolve_artifact_dir_env()) / _AGENT_WORKSPACE_DIR
db_path = os.environ.get("SYNTHORG_DB_PATH", "").strip()
if db_path:
return Path(db_path).parent / _AGENT_WORKSPACE_DIR
if os.environ.get("SYNTHORG_DATABASE_URL", "").strip():
return Path(_DEFAULT_DATA_DIR) / _AGENT_WORKSPACE_DIR
return NoneReferences
- Avoid hardcoded values; define them as constants. (link)
| """ | ||
| if self._agent_workspace_root is not None: | ||
| return self._agent_workspace_root | ||
| return Path(tempfile.gettempdir()) / "synthorg-agent-workspaces" |
There was a problem hiding this comment.
The string "synthorg-agent-workspaces" is hardcoded here. It should be defined as a constant (e.g., _DEFAULT_WORKSPACE_DIR) to comply with the project's "no-hardcoded-values" policy.
References
- Avoid hardcoded values; define them as constants. (link)
| async def _resolve_identity( | ||
| self, | ||
| assigned_to: str | None, | ||
| *, | ||
| task_id: str, | ||
| ) -> AgentIdentity: | ||
| """Resolve the task's assigned agent identity, or raise.""" | ||
| if assigned_to: | ||
| identity = await self._agent_registry.get(assigned_to) | ||
| if identity is None: | ||
| identity = await self._agent_registry.get_by_name(assigned_to) | ||
| if identity is not None: | ||
| return identity | ||
| logger.warning( | ||
| WORKERS_EXECUTION_SERVICE_NO_OP, | ||
| task_id=task_id, | ||
| reason="agent_not_registered", | ||
| assigned_to=assigned_to, | ||
| ) | ||
| msg = ( | ||
| f"Task {task_id!r} is assigned to an agent that is not " | ||
| f"registered in the runtime. Complete setup so agents are " | ||
| f"bootstrapped before submitting work." | ||
| ) | ||
| raise AgentRuntimeNotConfiguredError(msg) |
There was a problem hiding this comment.
When assigned_to is None, the error message and exception type are misleading. AgentRuntimeNotConfiguredError is intended for missing LLM providers, but an unassigned task is a state issue for that specific task. The current message also incorrectly states the task is assigned to an agent that is not registered when it is actually not assigned at all. Handling the None case separately with a more accurate error message and type (e.g., NotFoundError) improves clarity.
async def _resolve_identity(
self,
assigned_to: str | None,
*,
task_id: str,
) -> AgentIdentity:
"""Resolve the task's assigned agent identity, or raise."""
if not assigned_to:
logger.warning(
WORKERS_EXECUTION_SERVICE_NO_OP,
task_id=task_id,
reason="task_unassigned",
)
msg = f"Task {task_id!r} is not assigned to any agent."
raise NotFoundError(msg)
identity = await self._agent_registry.get(assigned_to)
if identity is None:
identity = await self._agent_registry.get_by_name(assigned_to)
if identity is not None:
return identity
logger.warning(
WORKERS_EXECUTION_SERVICE_NO_OP,
task_id=task_id,
reason="agent_not_registered",
assigned_to=assigned_to,
)
msg = (
f"Task {task_id!r} is assigned to agent {assigned_to!r} which is not "
f"registered in the runtime. Complete setup so agents are "
f"bootstrapped before submitting work."
)
raise AgentRuntimeNotConfiguredError(msg)| autonomy_resolver = AutonomyResolver( | ||
| registry=ActionTypeRegistry(), | ||
| config=AutonomyConfig(), | ||
| ) |
There was a problem hiding this comment.
The AutonomyResolver is being initialized with a default AutonomyConfig(). This likely ignores any autonomy-related settings defined in the application configuration (app_state.config.security). The resolver should be populated from the application's configuration to ensure user settings are respected.
| autonomy_resolver = AutonomyResolver( | |
| registry=ActionTypeRegistry(), | |
| config=AutonomyConfig(), | |
| ) | |
| autonomy_resolver = AutonomyResolver( | |
| registry=ActionTypeRegistry(), | |
| config=app_state.config.security.autonomy, | |
| ) |
Aureliolo
temporarily deployed
to
cloudflare-preview
GitHub Actions
Inactive
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/synthorg/api/state.py`:
- Around line 930-936: set_agent_workspace_root currently accepts and persists
relative Paths which can cause sandbox writes relative to the current working
directory; convert and persist an absolute Path instead: inside
set_agent_workspace_root, if the provided path is not absolute call
path.resolve(strict=False) (or Path.cwd()/path if you prefer) to produce an
absolute Path, then pass that absolute Path into the existing _set_once call so
_agent_workspace_root is always stored as an absolute path.
In `@src/synthorg/workers/execution_service.py`:
- Around line 299-306: The code wrongly raises NotFoundError when a task exists
but is unassigned; replace the NotFoundError in the assigned_to check with a
domain/state error (e.g., raise InvalidTaskStateError or TaskStateError) so
callers treat this as a configuration/scheduling problem, not a 404; keep the
existing logger.warning call (WORKERS_EXECUTION_SERVICE_NO_OP, task_id=task_id,
reason="task_unassigned") and raise the new exception with the same descriptive
message (f"Task {task_id!r} is not assigned to any agent.") and ensure you
import or define the chosen exception type in the module's exceptions import
list.
- Around line 260-264: Wrap the call to self._engine.run(...) in a try/except
that catches Exception, logs an ERROR (or WARNING as appropriate) including
task_id and agent_id (pull them from the local task/identity objects or their
attributes), the effective_autonomy and any short context, then re-raise the
exception; update the block around self._engine.run (where run_result is
assigned) to ensure failures are logged with task_id/agent_id before
propagating.
In `@tests/unit/providers/drivers/test_scripted.py`:
- Around line 126-129: Replace the loose assertion in test_warns_on_construction
so it looks for the specific safety-warning emitted by ScriptedDriver instead of
any warning: call ScriptedDriver(strategy=DeterministicResponseStrategy())
inside capture_logs() and then assert that there exists a log entry where
entry.get("log_level") == "warning" and the entry message (e.g.,
entry.get("message","")) contains the expected safety text or stable identifier
(for example the substring "ScriptedDriver" or the exact safety-warning string
your code emits); this pins the test to the specific safety-log contract
produced by ScriptedDriver rather than any unrelated warning.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI (base), Organization UI (inherited)
Review profile: ASSERTIVE
Plan: Pro
Run ID: b377d445-2abd-4e9e-a737-3a47d48451ae
📒 Files selected for processing (10)
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.pysrc/synthorg/workers/execution_service.pysrc/synthorg/workers/runtime_builder.pytests/e2e/test_runtime_online_seam.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/controllers/test_setup.pytests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/providers/drivers/test_scripted.py
📜 Review details
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (8)
- GitHub Check: Build Backend
- GitHub Check: Dashboard Test
- GitHub Check: Test Integration
- GitHub Check: Test E2E
- GitHub Check: Test Unit
- GitHub Check: Build Web Assets (melange)
- GitHub Check: Lighthouse Dashboard
- GitHub Check: Analyze (python)
🧰 Additional context used
📓 Path-based instructions (7)
**/*.{py,ts,tsx,js,jsx}
📄 CodeRabbit inference engine (CLAUDE.md)
No region/currency/locale privileged; use metric units; use British English per docs/reference/regional-defaults.md
Files:
tests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pysrc/synthorg/api/app_helpers.pytests/e2e/test_runtime_online_seam.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/controllers/test_setup.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pytests/unit/providers/drivers/test_scripted.pysrc/synthorg/workers/execution_service.py
tests/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Test markers:
@pytest.mark.{unit,integration,e2e,slow}; asyncauto; timeout 30s global; coverage 80% minWindows: unit tests use
WindowsSelectorEventLoopPolicy(3.14 IOCP teardown race); subprocess tests override backTest doubles:
FakeClockfor Clock seam,mock_of[T](**overrides)for typed-boundary substitutions,SimpleNamespacefor attribute-bags. BareMagicMockat typed boundaries is blocked byscripts/check_mock_spec.py(zero-tolerance, no baseline)FakeClock and
mock_ofimport fromtests._shared; inject viaclock=and the helper's spec subscriptHypothesis: 10 deterministic CI examples; failures are real bugs (fix + add
@example(...))Flaky tests: NEVER skip/xfail; fix fundamentally. Use
asyncio.Event().wait()notsleep(large)
Files:
tests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/e2e/test_runtime_online_seam.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/controllers/test_setup.pytests/unit/providers/drivers/test_scripted.py
⚙️ CodeRabbit configuration file
Test files do not require Google-style docstrings on classes or functions -- ruff D rules are only enforced on src/. A bare
@settings() decorator with no arguments on Hypothesis property tests is a no-op and should not be suggested -- the HYPOTHESIS_PROFILE env var controls example counts via registered profiles, which@given() honors automatically.
Files:
tests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/e2e/test_runtime_online_seam.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/controllers/test_setup.pytests/unit/providers/drivers/test_scripted.py
src/synthorg/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Configuration precedence: DB > env > code default via
SettingsService/ConfigResolver(Cat-1) or env > code default (Cat-2,read_only_post_init); Cat-3 bootstrap secrets are pure env; noos.environ.getoutside startup per docs/reference/configuration-precedence.mdNumerics must live in
settings/definitions/; allow 0/1/-1, HTTP codes, hex masks, powers-of-2, and module-level annotated named constants (NAME: int|float|Final|Final[int]|Final[float] = literal). Enforced byscripts/check_no_magic_numbers.py
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
src/**/*.{py,ts,tsx,js,jsx}
📄 CodeRabbit inference engine (CLAUDE.md)
Comments must document WHY only; no reviewer citations / issue back-refs / migration framing. Enforced by
check_no_review_origin_in_code.py+check_no_migration_framing.pyVendor-agnostic names in project code/tests: use
example-provider,test-provider,example-{large,medium,small}-001instead of real vendor names. Allowed in.claude/, third-party imports,providers/presets.py,web/public/provider-logos/
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
src/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
No
from __future__ import annotationsin Python 3.14+; use PEP 758 exceptexcept A, B:(parens required for binding)Type hints on public functions; mypy strict; Google-style docstrings; line length 88; functions <50 lines; files <800 lines
Errors must use
<Domain><Condition>ErrorfromDomainError; never inheritException/RuntimeErrordirectly. Enforced bycheck_domain_error_hierarchy.pyPydantic v2 frozen +
extra="forbid"on every frozen model project-wide (gatecheck_frozen_model_extra_forbid.py;@computed_fieldauto-exempt, per-line# lint-allow: frozen-extra-forbid -- <reason>forextra="allow"/"ignore"boundaries)Use
@computed_fieldfor derived model fields;NotBlankStrfor identifiersArgs models at every system boundary;
parse_typed()for every external dict ingestion. Enforced bycheck_boundary_typed.pyImmutability: use
model_copy(update=...)orcopy.deepcopy(); deepcopy at system boundariesAsync: use
asyncio.TaskGroupfor fan-out/fan-in; helpers catchException(re-raiseMemoryError/RecursionError)Clock seam: inject
clock: Clock | None = Noneparameter; tests injectFakeClock; lifecycle: services own_lifecycle_lock; timed-out stops mark unrestartableUntrusted content (SEC-1): use
wrap_untrusted()fromengine.prompt_safety; useHTMLParseGuardfor HTML processingLogging: use
from synthorg.observability import get_logger; variable name alwayslogger; never useimport loggingorprint()in app codeEvent names from
observability.events.<domain>constants; use structured kwargs (logger.info(EVENT, key=value))Error paths log WARNING/ERROR with context before raising; state transitions log INFO via
*_STATUS_TRANSITIONEDAFTER persistence writeSecret-log redaction (SEC-1): never use
error=str(exc)or interpolate{exc}; useerror_type=type(exc).__name__+error=safe_error_description(exc); never useexc_info=True; OTel: use `span.set_attribute("exception.message",...
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
⚙️ CodeRabbit configuration file
This project uses Python 3.14+ with PEP 758 except syntax: "except A, B:" (comma-separated, no parentheses) is correct and mandatory -- do NOT flag it as a typo or suggest parenthesized form. The "except builtins.MemoryError, RecursionError: raise" pattern is intentional project convention for system-error propagation. When evaluating the 50-line function limit, count only the function body excluding the signature lines, decorators, and docstring. Functions 1-5 lines over due to docstrings or multi-line signatures should not be flagged. Do not suggest extracting single-use helper functions called exactly once -- this reduces readability without improving maintainability.
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
src/synthorg/api/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
API startup lifecycle: two phases - construction (synchronous services in
create_appbody) and on_startup (services needing persistence). Ordering invariants:agent_registrybeforeauto_wire_meetings;tunnel_providerunconditional; SettingsService beforeWorkflowExecutionObserver;OntologyServiceafterpersistence.connect()
Files:
src/synthorg/api/app_helpers.pysrc/synthorg/api/state.py
src/synthorg/workers/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Worker execution service:
synthorg.workers.runtime_builder.build_worker_execution_serviceselects provider-backed or empty-company backstop; install viaAppState.worker_execution_serviceseam; empty-company rejects withAgentRuntimeNotConfiguredError(4014); useswap_worker_execution_service/swap_provider_registrywith lock synchronisation
Files:
src/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
🧠 Learnings (2)
📓 Common learnings
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Read `docs/design/` page before implementing; deviations require approval per DESIGN_SPEC.md
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Present every plan for accept/deny before coding (Planning gate)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Every convention PR must ship its enforcement gate per docs/reference/convention-gates.md
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: After issue: branch + commit + push (no auto-PR); use `/pre-pr-review`; no `/aurelio-review-pr` for external feedback until after PR creation. Fix EVERYTHING valid; no deferring.
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: xdist `-n 8 --dist=loadfile` auto-applied via pyproject `addopts` (`loadfile` prevents 3.14+Windows ProactorEventLoop leak)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Commits: `<type>: <description>` (feat/fix/refactor/docs/test/chore/perf/ci); commitizen-enforced
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Signed commits required on protected refs (GPG/SSH or GitHub App via `synthorg-repo-bot`)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Branches: `<type>/<slug>` from main
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Pre-commit/pre-push hooks: `.pre-commit-config.yaml`; tool-call gates: `.claude/settings.json` PreToolUse (`scripts/check_*.sh`/`.py`)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Squash merge; PR body becomes squash commit; trailers (`Release-As`, `Closes `#N``) must be in PR body
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: GitHub queries: use `gh issue list` via Bash, NOT MCP `list_issues`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: After every squash merge → `/post-merge-cleanup`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: CLI is Docker-only (init/start/stop/status); features go in dashboard + REST API
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: See `web/CLAUDE.md` for Web framework conventions (React 19); see `cli/CLAUDE.md` for CLI (Go binary) conventions
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T06:43:06.946Z
Learning: Python 3.14+ only; BUSL-1.1 → Apache 2.0 after Change Date; layout: `src/synthorg/` (src layout), `tests/`, `web/` (React 19), `cli/` (Go binary)
📚 Learning: 2026-05-05T09:04:46.195Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1760
File: scripts/_dual_backend_parity_lib.py:215-216
Timestamp: 2026-05-05T09:04:46.195Z
Learning: This repository targets Python 3.14+ and follows PEP 758. Therefore, reviewer tooling should NOT treat unparenthesized multi-exception `except` clauses written without an `as` clause (e.g., `except MemoryError, RecursionError:`) as syntax errors. Only flag `except`-clause problems when they are genuinely invalid for Python 3.14+.
Applied to files:
tests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pysrc/synthorg/api/app_helpers.pytests/e2e/test_runtime_online_seam.pytests/integration/api/test_task_create_empty_company.pytests/unit/api/controllers/test_setup.pysrc/synthorg/api/state.pysrc/synthorg/workers/runtime_builder.pytests/unit/providers/drivers/test_scripted.pysrc/synthorg/workers/execution_service.py
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #2003 +/- ##
==========================================
- Coverage 84.97% 84.97% -0.01%
==========================================
Files 1880 1882 +2
Lines 110872 111136 +264
Branches 9457 9481 +24
==========================================
+ Hits 94218 94440 +222
- Misses 14343 14372 +29
- Partials 2311 2324 +13 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
CI: the new shared session-scoped provider_registry fixture poisoned the shared app, breaking two pre-existing no-provider assertions. Both tests now swap an empty ProviderRegistry and restore in finally (mirrors test_complete_succeeds_with_all_prerequisites): test_complete_rejects_without_providers and test_returns_status_with_seeded_users. CodeRabbit: app_helpers validates SYNTHORG_DB_PATH is absolute before deriving the workspace root (raises, consistent with the artifact-dir guard); test_runtime_online_seam tmp_path typed as Path with the type-ignore dropped; test_task_create_empty_company matches Set-Cookie case-insensitively; test_state_swap asserts ServiceUnavailableError instead of bare Exception; removed redundant in-test structlog.reset_defaults() in two tests since the autouse _reset_structlog_state fixture already isolates. Gemini: app_helpers and state hoist the agent-workspaces and data-dir literals to module Final constants; execution_service raises NotFoundError with an accurate message for an unassigned task instead of AgentRuntimeNotConfiguredError; runtime_builder feeds AutonomyResolver the configured app_state.config.config.autonomy instead of a default AutonomyConfig().
All four are CodeRabbit follow-ups on the round-1 fixes, reviewed on the new head. set_agent_workspace_root rejects non-absolute paths (ValueError), matching the resolve_agent_workspace_root_env guard so sandbox writes cannot land cwd-relative. engine.run() is now wrapped so a provider/tooling failure logs an ERROR with task_id and agent_id before propagating (WORKERS_EXECUTION_SERVICE_FAILED). The existing-but-unassigned task now raises ConflictError (409) instead of NotFoundError: this reconciles the round-1 Gemini ask (do not use the misleading AgentRuntimeNotConfiguredError) with CodeRabbit round 2 (a loaded task is not a 404) -- a 409 state conflict is accurate for both. test_warns_on_construction now pins the specific PROVIDER_SCRIPTED_DRIVER_INSTANTIATED event and strategy field instead of any warning. Added coverage for every new path: test_set_rejects_relative_path, test_unassigned_task_raises_conflict, test_engine_run_failure_logs_and_reraises.
Aureliolo
force-pushed
the
fix/agent-runtime-online
branch
from
6829f3a to
f745bd4
Compare
Aureliolo
temporarily deployed
to
cloudflare-preview
GitHub Actions
Inactive
There was a problem hiding this comment.
Actionable comments posted: 5
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/synthorg/workers/runtime_builder.py`:
- Around line 42-143: The build_worker_execution_service function is too large;
split its responsibilities into small helpers: extract provider resolution into
a helper like resolve_provider(registry) that checks
app_state.has_active_provider, calls registry.list_providers(), logs warnings
and returns the selected provider or raises/returns None; extract workspace
creation and web timeout/config read into
prepare_workspace_and_timeout(workspace_root, config_resolver) which ensures
workspace_root.mkdir(...) and returns web_request_timeout; extract tool
construction into build_tool_registry(workspace_root, app_state.config,
web_request_timeout) which calls build_default_tools_from_config and returns a
ToolRegistry; extract engine construction into construct_agent_engine(provider,
registry, tool_registry, app_state) that builds and returns the AgentEngine
instance; extract autonomy resolver creation into
create_autonomy_resolver(app_state.config); then have
build_worker_execution_service call these helpers in sequence, log the final
info, and return AgentEngineExecutionService(agent_engine, task_engine,
agent_registry, autonomy_resolver).
In `@tests/_shared/scripted_provider.py`:
- Around line 75-92: The constructor __init__ currently accepts conflicting
inputs (responses, response, error) and silently prioritizes responses; update
__init__ to validate that at most one of responses, response, or error is
provided and raise a ValueError if multiple are supplied, then keep the existing
logic that sets self._strategy to SequencedResponseStrategy or
SingleResponseStrategy (or None) and preserve deepcopy of capabilities
(TEST_CAPABILITIES) and assignment to self._capabilities and self._strategy.
- Around line 152-155: get_model_capabilities currently returns the internal
mutable _capabilities object by reference which allows tests to mutate shared
state; change get_model_capabilities to return a defensive copy instead (e.g.,
use copy.deepcopy(self._capabilities) or construct a new ModelCapabilities from
the fields of self._capabilities) so callers get an independent object; add the
needed import (copy) if using deepcopy and ensure the async method returns the
copied instance rather than self._capabilities.
In `@tests/integration/api/test_task_create_empty_company.py`:
- Around line 44-46: After calling _extract_auth_cookies(...) store and validate
the returned session_token and csrf_token immediately: ensure neither is
empty/None and raise/assert with a clear message so the test fails fast on
bootstrap regressions; update the test in test_task_create_empty_company.py
where session_token and csrf_token are set (and before setting client.headers
and X-CSRF-Token) to perform these checks.
In `@tests/unit/workers/test_runtime_builder.py`:
- Line 11: Replace the SystemClock import with the shared FakeClock and mock_of
from tests._shared, then instantiate/use FakeClock in the tests instead of
SystemClock and inject it via clock= parameters where relevant; specifically
change the top-level import "from synthorg.core.clock import SystemClock" to
"from tests._shared import FakeClock, mock_of", replace any SystemClock() usages
with FakeClock(), and where helpers use mocked types switch to the helper's spec
subscript form using mock_of[YourType] (apply same changes for the other
occurrences mentioned around lines 69-70, 102-103, 134-135).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI (base), Organization UI (inherited)
Review profile: ASSERTIVE
Plan: Pro
Run ID: dbba6097-1b89-46e9-90d3-c245a31b0b1d
📒 Files selected for processing (35)
CLAUDE.mdscripts/_ghost_wiring_manifest.txtsrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/api/state.pysrc/synthorg/api/state_services.pysrc/synthorg/core/domain_errors.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/provider.pysrc/synthorg/observability/events/workers.pysrc/synthorg/providers/drivers/scripted.pysrc/synthorg/providers/registry.pysrc/synthorg/workers/execution_service.pysrc/synthorg/workers/runtime_builder.pytests/_shared/scripted_provider.pytests/e2e/conftest.pytests/e2e/test_runtime_online_seam.pytests/integration/api/conftest.pytests/integration/api/test_runtime_install_ordering.pytests/integration/api/test_task_create_empty_company.pytests/integration/engine/test_coordination_wiring.pytests/unit/api/conftest.pytests/unit/api/controllers/test_coordination.pytests/unit/api/controllers/test_setup.pytests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/engine/quality/scripted_provider.pytests/unit/providers/drivers/test_scripted.pytests/unit/workers/test_execution_service.pytests/unit/workers/test_runtime_builder.pyweb/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
📜 Review details
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (7)
- GitHub Check: Build Web Assets (melange)
- GitHub Check: Dashboard Test
- GitHub Check: Test Unit
- GitHub Check: Test E2E
- GitHub Check: Test Integration
- GitHub Check: Lighthouse Dashboard
- GitHub Check: Analyze (python)
🧰 Additional context used
📓 Path-based instructions (12)
src/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Use
SettingsService/ConfigResolverfor DB > env > code default (Cat-1) or env > code default (Cat-2,read_only_post_init); Cat-3 bootstrap secrets are pure env at the boot site; noos.environ.getoutside startup; pre-init Cat-2 reads usesettings.bootstrap_resolver.resolve_init_valueper docs/reference/configuration-precedence.mdNo
from __future__ import annotations(3.14 has PEP 649); use PEP 758 except:except A, B:no parens unless bindingType hints on public functions; mypy strict; Google-style docstrings; line length 88; functions <50 lines; files <800 lines
Errors must follow
<Domain><Condition>Errornaming fromDomainError; never inheritException/RuntimeError/etc directly; enforced bycheck_domain_error_hierarchy.pyPydantic v2 frozen +
extra="forbid"on every frozen model project-wide; gatecheck_frozen_model_extra_forbid.py;@computed_fieldauto-exempt, per-line# lint-allow: frozen-extra-forbid -- <reason>forextra="allow"/"ignore"boundaries; use@computed_fieldfor derived fieldsUse
NotBlankStrfor identifiers in Pydantic modelsArgs models at every system boundary;
parse_typed()for every external dict ingestion; enforced bycheck_boundary_typed.pyUse immutability patterns:
model_copy(update=...)orcopy.deepcopy(); deepcopy at system boundariesUse
asyncio.TaskGroupfor fan-out/fan-in; helpers catchException(re-raiseMemoryError/RecursionError)Clock seam:
clock: Clock | None = None; tests injectFakeClock; lifecycle services own_lifecycle_lock; timed-out stops mark unrestartableUntrusted content (SEC-1): use
wrap_untrusted()fromengine.prompt_safety; useHTMLParseGuardfor HTMLUse
from synthorg.observability import get_logger; variable always namedlogger; neverimport logging/print()in app codeEvent names from
observability.events.<domain>constants; use structured kwargs (`logger.info(EVENT, key=valu...
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/core/domain_errors.pysrc/synthorg/providers/registry.pysrc/synthorg/api/state_services.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.pysrc/synthorg/api/state.pysrc/synthorg/providers/drivers/scripted.py
⚙️ CodeRabbit configuration file
This project uses Python 3.14+ with PEP 758 except syntax: "except A, B:" (comma-separated, no parentheses) is correct and mandatory -- do NOT flag it as a typo or suggest parenthesized form. The "except builtins.MemoryError, RecursionError: raise" pattern is intentional project convention for system-error propagation. When evaluating the 50-line function limit, count only the function body excluding the signature lines, decorators, and docstring. Functions 1-5 lines over due to docstrings or multi-line signatures should not be flagged. Do not suggest extracting single-use helper functions called exactly once -- this reduces readability without improving maintainability.
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/core/domain_errors.pysrc/synthorg/providers/registry.pysrc/synthorg/api/state_services.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.pysrc/synthorg/api/state.pysrc/synthorg/providers/drivers/scripted.py
src/synthorg/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Numerics live in
settings/definitions/; allowlist only 0/1/-1, HTTP codes, hex masks, powers-of-2, and module-level annotated named constants of the formNAME: int|float|Final|Final[int]|Final[float] = literal; enforced byscripts/check_no_magic_numbers.py
Files:
src/synthorg/observability/events/provider.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/error_taxonomy.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/core/domain_errors.pysrc/synthorg/providers/registry.pysrc/synthorg/api/state_services.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.pysrc/synthorg/api/state.pysrc/synthorg/providers/drivers/scripted.py
{src,web}/**/*.{py,ts,tsx,js,jsx}
📄 CodeRabbit inference engine (CLAUDE.md)
Comments must explain WHY only; no reviewer citations / issue back-refs / migration framing; enforced by
check_no_review_origin_in_code.py+check_no_migration_framing.py
Files:
src/synthorg/observability/events/provider.pyweb/src/api/types/error-codes.gen.tssrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/error_taxonomy.pyweb/src/api/types/openapi.gen.tssrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/core/domain_errors.pysrc/synthorg/providers/registry.pysrc/synthorg/api/state_services.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.pysrc/synthorg/api/state.pysrc/synthorg/providers/drivers/scripted.py
web/src/**/*.{js,jsx,ts,tsx,mts}
📄 CodeRabbit inference engine (web/CLAUDE.md)
web/src/**/*.{js,jsx,ts,tsx,mts}: Always usecreateLoggerfrom@/lib/logger; never bareconsole.warn/console.error/console.debugin application code. Variable name must always belog. Onlylogger.tsitself may use bare console methods. Uselog.debug()(DEV-only, stripped in production),log.warn(),log.error().
Pass dynamic/untrusted values as separate args to logger calls (not interpolated into the message string) so they go throughsanitizeArg
Attacker-controlled fields inside structured objects must be wrapped insanitizeForLog()before embedding in log calls
Error-code constants (MANDATORY): importErrorCodeandErrorCategoryfrom@/api/types/errors(re-exported from the generatedweb/src/api/types/error-codes.gen.ts). Discriminate onErrorCode.<NAME>, never on raw integer literals.
Use@eslint-react/web-api-no-leaked-fetchto detectfetch()in effects withoutAbortControllercleanup
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/api/types/**/*.gen.ts
📄 CodeRabbit inference engine (web/CLAUDE.md)
Generated DTO types (MANDATORY): NEVER hand-edit
web/src/api/types/*.gen.ts. Regenerate withuv run python scripts/generate_dto_types_ts.py. Import DTOs via the barrel (import type { AgentConfig } from '@/api/types').
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/**/*.{ts,tsx,mts}
📄 CodeRabbit inference engine (web/CLAUDE.md)
web/src/**/*.{ts,tsx,mts}: Use@typescript-eslint/no-floating-promisesto forbid unawaited promises so async work cannot survive the test that scheduled it and trip the active-handle gate
Use@typescript-eslint/no-misused-promises(withchecksVoidReturn: { attributes: false }) to forbid passing async functions where the callsite ignores the returned promise. React 19asyncevent handlers stay allowed via theattributes: falseexemption.
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
web/src/**/*.{ts,tsx}
📄 CodeRabbit inference engine (CLAUDE.md)
Reuse
web/src/components/ui/design tokens only in Web Dashboard Design System perweb/CLAUDE.md
Files:
web/src/api/types/error-codes.gen.tsweb/src/api/types/openapi.gen.ts
tests/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Test markers:
@pytest.mark.{unit,integration,e2e,slow}; asyncauto; timeout 30s global; coverage 80% minWindows: unit tests use
WindowsSelectorEventLoopPolicy(3.14 IOCP teardown race); subprocess tests override backTest doubles: use ladder in conventions.md section 12.1;
FakeClockfor Clock seam,mock_of[T](**overrides)for typed-boundary substitutions,SimpleNamespacefor attribute-bags; bareMagicMockat typed boundary is blocked byscripts/check_mock_spec.pyFakeClock and
mock_ofimport fromtests._shared; inject viaclock=and helper's spec subscriptHypothesis: 10 deterministic CI examples; failures are real bugs (fix + add
@example(...))Flaky tests: NEVER skip/xfail; fix fundamentally; use
asyncio.Event().wait()notsleep(large)
Files:
tests/unit/workers/test_runtime_builder.pytests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/api/controllers/test_coordination.pytests/integration/engine/test_coordination_wiring.pytests/integration/api/test_runtime_install_ordering.pytests/e2e/test_runtime_online_seam.pytests/unit/api/controllers/test_setup.pytests/unit/engine/quality/scripted_provider.pytests/unit/providers/drivers/test_scripted.pytests/unit/api/conftest.pytests/e2e/conftest.pytests/integration/api/test_task_create_empty_company.pytests/unit/workers/test_execution_service.pytests/integration/api/conftest.pytests/_shared/scripted_provider.py
⚙️ CodeRabbit configuration file
Test files do not require Google-style docstrings on classes or functions -- ruff D rules are only enforced on src/. A bare
@settings() decorator with no arguments on Hypothesis property tests is a no-op and should not be suggested -- the HYPOTHESIS_PROFILE env var controls example counts via registered profiles, which@given() honors automatically.
Files:
tests/unit/workers/test_runtime_builder.pytests/unit/api/controllers/test_setup_status.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/api/controllers/test_coordination.pytests/integration/engine/test_coordination_wiring.pytests/integration/api/test_runtime_install_ordering.pytests/e2e/test_runtime_online_seam.pytests/unit/api/controllers/test_setup.pytests/unit/engine/quality/scripted_provider.pytests/unit/providers/drivers/test_scripted.pytests/unit/api/conftest.pytests/e2e/conftest.pytests/integration/api/test_task_create_empty_company.pytests/unit/workers/test_execution_service.pytests/integration/api/conftest.pytests/_shared/scripted_provider.py
src/synthorg/api/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
API startup lifecycle: construction phase (
create_appbody) wires synchronous services; on_startup phase (_build_lifecycle.on_startup) wires services needing persistence; ordering invariants:agent_registrybeforeauto_wire_meetings;tunnel_providerwired unconditionally; SettingsService beforeWorkflowExecutionObserver;OntologyServiceafterpersistence.connect()
Files:
src/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pysrc/synthorg/api/state_services.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/api/state.py
src/synthorg/api/controllers/setup/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Setup completion:
post_setup_reinit()propagates failures;settings_svc.set("api", "setup_complete", "true")only after reinit succeeds; serialize check/validate/reinit/persist underCOMPLETE_LOCKto prevent race conditions
Files:
src/synthorg/api/controllers/setup/agent_helpers.py
src/synthorg/providers/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Resilience: provider calls through
BaseCompletionProvider(retry + rate limit); never implement retry in driver subclasses; retryable:RateLimitError,Provider{Timeout,Connection,Internal}Error
Files:
src/synthorg/providers/registry.pysrc/synthorg/providers/drivers/scripted.py
src/synthorg/workers/**/*.py
📄 CodeRabbit inference engine (CLAUDE.md)
Worker execution service: select via
synthorg.workers.runtime_builder.build_worker_execution_servicebehind provider-present switch; install viaAppState.worker_execution_serviceseam; append boot install hook FIRST after persistence/SettingsService; empty-company rejects task creation withAgentRuntimeNotConfiguredError(4014); useswap_worker_execution_service/swap_provider_registrywith lock synchronisation
Files:
src/synthorg/workers/runtime_builder.pysrc/synthorg/workers/execution_service.py
🧠 Learnings (4)
📓 Common learnings
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Read `docs/design/` page before implementing; deviations need approval per [DESIGN_SPEC.md](docs/DESIGN_SPEC.md)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Present every plan for accept/deny before coding
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: No region/currency/locale privileged; use metric units; use British English per [docs/reference/regional-defaults.md](docs/reference/regional-defaults.md)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Every convention PR ships its enforcement gate per [docs/reference/convention-gates.md](docs/reference/convention-gates.md)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Never edit `tests/baselines/unit_timing.json` or any `scripts/*_baseline.{txt,json}` / `scripts/_*_baseline.py`; timeout/slow failures = source-code regression; both families are PreToolUse-blocked; per-invocation bypass requires explicit user approval via `ALLOW_BASELINE_GROWTH=1 git commit ...`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: After issue: branch + commit + push (no auto-PR); use `/pre-pr-review`; after PR use `/aurelio-review-pr` for external feedback; fix EVERYTHING valid; no deferring
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Commits: `<type>: <description>` (feat/fix/refactor/docs/test/chore/perf/ci); commitizen-enforced
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Signed commits required on protected refs (GPG/SSH or GitHub App via `synthorg-repo-bot`)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Branches: `<type>/<slug>` from main
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Use `.pre-commit-config.yaml` for pre-commit/pre-push hooks; tool-call gates: `.claude/settings.json` PreToolUse (`scripts/check_*.sh`/`.py`)
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: Squash merge; PR body becomes squash commit; trailers (`Release-As`, `Closes `#N``) must be in PR body
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: GitHub queries: use `gh issue list` via Bash, NOT MCP `list_issues`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: After every squash merge → `/post-merge-cleanup`
Learnt from: CR
Repo: Aureliolo/synthorg
Timestamp: 2026-05-18T07:15:31.194Z
Learning: CLI is Docker-only (init/start/stop/status); features go in dashboard + REST API
📚 Learning: 2026-05-05T09:04:46.195Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1760
File: scripts/_dual_backend_parity_lib.py:215-216
Timestamp: 2026-05-05T09:04:46.195Z
Learning: This repository targets Python 3.14+ and follows PEP 758. Therefore, reviewer tooling should NOT treat unparenthesized multi-exception `except` clauses written without an `as` clause (e.g., `except MemoryError, RecursionError:`) as syntax errors. Only flag `except`-clause problems when they are genuinely invalid for Python 3.14+.
Applied to files:
src/synthorg/observability/events/provider.pytests/unit/workers/test_runtime_builder.pytests/unit/api/controllers/test_setup_status.pysrc/synthorg/observability/events/api.pysrc/synthorg/observability/events/workers.pysrc/synthorg/core/error_taxonomy.pytests/unit/api/test_state_swap_worker_execution_service.pytests/unit/api/controllers/test_coordination.pysrc/synthorg/api/controllers/setup/agent_helpers.pysrc/synthorg/api/controllers/tasks.pytests/integration/engine/test_coordination_wiring.pysrc/synthorg/core/domain_errors.pysrc/synthorg/providers/registry.pysrc/synthorg/api/state_services.pytests/integration/api/test_runtime_install_ordering.pysrc/synthorg/api/app.pysrc/synthorg/api/app_helpers.pysrc/synthorg/workers/runtime_builder.pytests/e2e/test_runtime_online_seam.pytests/unit/api/controllers/test_setup.pytests/unit/engine/quality/scripted_provider.pytests/unit/providers/drivers/test_scripted.pysrc/synthorg/workers/execution_service.pytests/unit/api/conftest.pytests/e2e/conftest.pytests/integration/api/test_task_create_empty_company.pytests/unit/workers/test_execution_service.pytests/integration/api/conftest.pysrc/synthorg/api/state.pysrc/synthorg/providers/drivers/scripted.pytests/_shared/scripted_provider.py
📚 Learning: 2026-05-16T18:36:31.446Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1944
File: docs/reference/conventions.md:787-789
Timestamp: 2026-05-16T18:36:31.446Z
Learning: In Aureliolo/synthorg, do not require adding `<!--RS:...-->` “Doc Numeric Claims (MANDATORY)” numeric macros for Python version numbers mentioned in documentation prose (e.g., “Python 3.14”, “Python 3.15”). The `scripts/check_doc_numeric_macros.py` gate only applies to `README.md`, `docs/index.md`, `docs/roadmap/index.md`, `docs/architecture/decisions.md`, and `docs/reference/convention-gates.md`, and it only flags digits adjacent to specific stat nouns (tests/providers/agents/stars/releases), not language version mentions like “Python 3.14”.
Applied to files:
CLAUDE.md
📚 Learning: 2026-05-16T18:36:35.250Z
Learnt from: Aureliolo
Repo: Aureliolo/synthorg PR: 1944
File: docs/getting_started.md:109-109
Timestamp: 2026-05-16T18:36:35.250Z
Learning: When reviewing Markdown in the synthorg repo, account for the CI gate `check_doc_numeric_macros.py`: it skips fenced code blocks entirely, and it only flags digits that are adjacent to these stat nouns: `tests`, `providers`, `agents`, `stars`, `releases`. Therefore, numeric examples such as CLI flag values (e.g., `--num-workers=4` in fenced bash blocks) and prose version numbers (e.g., `3.14`/`3.15`) are not expected to trigger this check; prioritize changes only when digits appear next to one of the listed nouns (e.g., “5 tests”, “10 stars”, etc.).
Applied to files:
CLAUDE.md
🔇 Additional comments (33)
src/synthorg/observability/events/provider.py (1)
9-9: LGTM!web/src/api/types/error-codes.gen.ts (1)
57-57: LGTM!tests/unit/api/controllers/test_setup_status.py (1)
12-12: LGTM!Also applies to: 26-39
scripts/_ghost_wiring_manifest.txt (1)
26-26: LGTM!src/synthorg/observability/events/api.py (1)
64-64: LGTM!src/synthorg/observability/events/workers.py (1)
72-82: LGTM!src/synthorg/core/error_taxonomy.py (1)
109-109: LGTM!CLAUDE.md (1)
82-83: LGTM!web/src/api/types/openapi.gen.ts (1)
7295-7295: LGTM!tests/unit/api/test_state_swap_worker_execution_service.py (1)
1-81: LGTM!tests/unit/api/controllers/test_coordination.py (1)
127-130: LGTM!Also applies to: 142-144
src/synthorg/api/controllers/setup/agent_helpers.py (1)
142-166: LGTM!src/synthorg/api/controllers/tasks.py (1)
25-25: LGTM!Also applies to: 37-37, 182-189
tests/integration/engine/test_coordination_wiring.py (1)
285-289: LGTM!Also applies to: 300-302
src/synthorg/core/domain_errors.py (1)
248-264: LGTM!src/synthorg/providers/registry.py (1)
141-146: LGTM!src/synthorg/api/state_services.py (1)
137-137: LGTM!Also applies to: 451-460, 471-490, 527-531
tests/integration/api/test_runtime_install_ordering.py (1)
1-57: LGTM!src/synthorg/api/app.py (1)
33-34: LGTM!Also applies to: 980-1029
src/synthorg/api/app_helpers.py (1)
48-50: LGTM!Also applies to: 110-110, 129-161
tests/e2e/test_runtime_online_seam.py (1)
1-197: LGTM!tests/unit/api/controllers/test_setup.py (1)
372-380: LGTM!Also applies to: 385-385
tests/unit/engine/quality/scripted_provider.py (1)
1-18: LGTM!tests/unit/providers/drivers/test_scripted.py (1)
1-138: LGTM!src/synthorg/workers/execution_service.py (1)
5-31: LGTM!Also applies to: 36-55, 71-77, 99-113, 139-147, 197-410
tests/unit/api/conftest.py (1)
30-31: LGTM!Also applies to: 46-47, 356-370, 426-427, 459-460
tests/e2e/conftest.py (1)
1-6: LGTM!Also applies to: 12-18, 23-29
tests/integration/api/test_task_create_empty_company.py (1)
1-34: LGTM!Also applies to: 50-112
tests/unit/workers/test_execution_service.py (1)
1-296: LGTM!tests/integration/api/conftest.py (1)
12-31: LGTM!Also applies to: 34-93
src/synthorg/api/state.py (1)
9-14: LGTM!Also applies to: 84-88, 160-161, 178-178, 430-443, 860-865, 883-950, 968-973
src/synthorg/providers/drivers/scripted.py (1)
1-293: LGTM!tests/_shared/scripted_provider.py (1)
1-74: LGTM!Also applies to: 93-151, 157-269
CodeRabbit follow-ups on the round-2 head. build_worker_execution_service split into _select_active_provider, _build_tool_registry, and _construct_agent_engine helpers so each unit is under the 50-line rule. ScriptedProvider test double now raises ValueError when more than one of responses/response/error is supplied, and get_model_capabilities returns a deepcopy so callers cannot mutate shared capabilities state. The empty-company integration test asserts the auth bootstrap session and csrf cookies are non-empty before reuse. test_runtime_builder uses FakeClock from tests._shared instead of SystemClock per the clock-seam convention. All gates green: mypy-affected, pytest-affected, ghost-wiring, function-size.
Aureliolo
temporarily deployed
to
cloudflare-preview
GitHub Actions
Inactive
Aureliolo
temporarily deployed
to
cloudflare-preview
GitHub Actions
Inactive
…harness) (#2006) ## Summary Wires the GHOST `IntakeEngine` into the shipped boot path so the client-simulation runtime comes online and becomes the deterministic e2e acceptance substrate (independent child of EPIC #1955; runtime root #2003 already merged). - New `src/synthorg/client/runtime_builder.py` (`build_client_simulation_runtime`): constructs the real `IntakeEngine` + a single-stage `ReviewPipeline` (`InternalReviewStage`) and returns a populated `ClientSimulationState`. Called from `create_app`'s construction phase when a `TaskEngine` is present, so `has_simulation_runtime` is true and the `/simulations` + `/requests` controllers register. - Pluggable intake: `IntakeConfig` discriminator + `build_intake_strategy()` factory (`direct` -> `DirectIntake`, no LLM; `agent` -> `AgentIntake`, LLM triage). New `simulations.intake_strategy` / `simulations.intake_model` settings, resolved at the boot site via the bootstrap resolver (env > default; `read_only_post_init`, matching the existing `simulations.task_timeout_seconds` pattern). A selected `agent` strategy that cannot be satisfied degrades to `direct` with a WARNING; a `direct` failure propagates. - Default `direct` strategy makes zero LLM calls, so the runtime comes online for an empty company. - Docs synced: `docs/design/client-simulation.md` (removed the "not wired" warning, added the Boot wiring section + factory-table row); `README.md` (intake engine is now exercised end to end). ## CORE acceptance `scripts/_ghost_wiring_manifest.txt` `IntakeEngine` line flipped `PENDING` -> `ENFORCED` in this PR; `scripts/check_no_ghost_wiring.py` passes (verified construction site under `src/synthorg/client/`, reachable from `app.py`). ## Bundled fix: pre-existing latent circular import Wiring `runtime_builder` surfaced a pre-existing cycle: `synthorg.client.*` was unimportable as a *first* import in a fresh interpreter (`budget.cost_record` -> `ontology` -> `persistence` -> `security` -> `providers` -> back into budget mid-init). Reproduced on unmodified `synthorg.client.runner`, so it is not introduced here. Root fix: PEP 562 lazy export of `OntologyService` / `OntologyEntityRepository` from `synthorg.ontology.__init__` (locked double-checked cache, mirroring the established `synthorg.tools.mcp` pattern). Public API unchanged: `from synthorg.ontology import OntologyService` still works, resolved on first access. Authorized by the maintainer as in-scope. ## Tests (four layers) - Seam (e2e): builder drives a real `ClientRequest` SUBMITTED -> TASK_CREATED against a real `TaskEngine`. - Boot-wired HTTP surface: no kwarg; `/capabilities` reports simulations/requests on, routes register. - Direct + scripted-agent harness: `SimulationRunner.run` driven directly against the boot-wired runtime, deterministic asserted metrics, zero LLM spend. - Factory/builder/settings/lazy-export unit tests; obsolete capability tests updated to the new reality. The deterministic harness assertion drives `SimulationRunner.run` directly rather than the HTTP poll-to-completion path: litestar `TestClient` does not progress fire-and-forget `asyncio.create_task` background tasks between sync requests, so a poll-based e2e would be flaky. The HTTP surface (routes register, 201 on start) is still covered separately. ## Review coverage Pre-PR review ran 23 agents (17 specialist + 6 audit mini-pass). All valid findings (V1-V5) and the user-approved pre-existing/outside-diff items (P1: `web/src/api/endpoints/clients.ts` migrated to the generated `@/api/types` barrel, fixing the `SimulationStatusResponse` / `acceptance_criteria` drift; P2: pre-raise logging on intake `ValueError` paths; P3: named tolerance constants, function-length refactor) implemented. Seven findings assessed invalid (agent misreads of PEP 758 / bootstrap-resolver semantics, gate-sanctioned `@computed_field` exemption, established-pattern conflicts) with rationale in the triage record. Gates green: ruff, mypy (3902 files), `check_no_ghost_wiring.py`, affected pytest, full `pre-commit run --all-files`; web type-check, lint, 3182 tests. Closes #1961
## Summary Wires the four GHOST governance subsystems from EPIC #1955 so they actually run on a live agent behind the provider-present switch (the runtime root #1956/#2003 only wired the minimal safety spine). - **Single boot `ApprovalGate` + real park/resume.** One gate constructed at boot in `lifecycle_builder._wire_approval_gate` (backed by the dual-backend `ParkedContextRepository`), attached to `AppState` and injected into `AgentEngine` so the engine parks and `/approvals` resumes on the *same* gate. New `AgentEngine.resume_parked_run` restores the parked `AgentContext`, injects the approval decision (D21; SYSTEM-message follow-up since the loop appends the escalated tool result before the park check), and continues `_execute`. `AgentEngineExecutionService.dispatch_resume` runs the resume as a tracked `BackgroundTaskRegistry` task off the approve/reject request path, drained on shutdown. - **Progressive trust enforced.** `TrustService` threaded into the tool-invoker seam; effective tool permissions are narrowed to the more restrictive of identity level and earned trust level (auto-initialised on first sight). `DISABLED` strategy is a no-op; switching strategy changes enforcement. - **Agent → SynthOrg-MCP self-consumer.** New `engine/mcp_self_consumer.py` (pluggable: StrEnum mode + frozen config + factory + safe `DISABLED` default) exposes SynthOrg's own MCP tools to a running agent, trust-scoped via `MCPToolScoper`, with `actor=identity` threaded so `require_admin_guardrails` fails closed on admin tools. - **AutonomyChangeStrategy wired.** Built at boot from `config.autonomy.change_strategy` (default `HUMAN_ONLY`), attached to `AppState`; the autonomy controller now enforces the D6 seniority rule, consults the strategy, and enqueues a real approval (replacing the always-deny stub). - **Anti-ghost lock:** `scripts/_ghost_wiring_manifest.txt` flips `ApprovalGate`, `TrustService`, `build_mcp_self_consumer`, `build_autonomy_change_strategy` to `ENFORCED` in this PR. - `docs/design/security.md` updated to current runtime-enforcement state. ## Test plan - New unit suites: `test_agent_engine_resume.py`, `test_agent_engine_trust.py`, `test_mcp_self_consumer.py`, `test_enforcement.py`; extended approval-gate / runtime-builder / execution-service / approvals-helper / autonomy / startup-wiring / guards tests; conversation-shape lock for the resume-injection invariant. - `uv run ruff check src/ tests/` + `ruff format` + `uv run mypy src/ tests/`: clean (3903 files). - Affected-tests gate: 29154 passed, 18 skipped. Full pre-commit + pre-push (ghost-wiring, dto-types-in-sync, ESLint, all convention gates) green. - Acceptance verified end-to-end at every seam: park → `/approvals` → resume on approve/reject; trust-strategy switch changes the permitted tool set; agent invokes a SynthOrg MCP read tool; admin tool fails closed without confirm/reason/actor; autonomy request routes through the strategy. ## Review coverage Pre-PR reviewed by 22 review + audit-mini-pass agents (issue-resolution-verifier: all 5 acceptance criteria RESOLVED; ghost-wiring: PASS). 11 valid findings triaged and **all implemented** (`_audit/pre-pr-review/triage.md`): fail-safe `_cleanup_parked` (no silent duplicate resume), louder no-parked-context path, `wrap_untrusted` for the injected decision reason, self-validating `AutonomyLevelRequest.reason` + regenerated TS types, MSW `reason` validation, `NamedTuple` trust result, logging coverage, function-length extraction, larger resume drain budget, tighter guard test, added tests. The ~20 "`except A, B:` syntax error" flags were dismissed as a confirmed PEP 758 false positive (CLAUDE.md-mandated; all gates green). Closes #1957
<!-- HIGHLIGHTS_START --> ## Highlights > _AI-generated summary (model: `openai/gpt-4.1-mini` via GitHub Models). Commit-based changelog below._ ### What you'll notice - Multi-agent coordination is now active immediately on startup for smoother operation. - Governance rules are fully enforced during use, ensuring compliance at all times. - Coordination metrics update live, giving real-time insights into system activity. - Review agents are now reliably processed, preventing silent drops in tasks. - Sandbox containers can be reused for agents and tasks, speeding up execution and reducing overhead. ### What's new - Agents support online runtime with a minimal safety framework to improve stability. - Recorded LLM interactions can be deterministically replayed at the provider interface. - Distributed path validation has been enhanced for more robust data routing. - A client-simulation runtime was added for end-to-end testing of the IntakeEngine. - A new work pipeline spine architecture has been introduced to streamline task processing. ### Under the hood - Infrastructure, Python, and web dependencies have all been updated to latest versions. - Updated apko lockfiles in the CI/CD pipeline improve build consistency. <!-- HIGHLIGHTS_END --> :robot: I have created a release *beep* *boop* --- ## [0.8.6](v0.8.5...v0.8.6) (2026-05-19) ### Features * agent runtime online + minimal safety spine (runtime root) ([#2003](#2003)) ([e5eef1a](e5eef1a)), closes [#1956](#1956) * deterministic recorded-LLM cassette replay at the provider chokepoint ([#2010](#2010)) ([cabf55d](cabf55d)) * distributed path validation + hardening ([#2011](#2011)) ([a382e4a](a382e4a)), closes [#1966](#1966) * wire IntakeEngine via boot client-simulation runtime (e2e test harness) ([#2006](#2006)) ([6a9c0aa](6a9c0aa)), closes [#1961](#1961) * work pipeline spine ([#1960](#1960)) ([#2013](#2013)) ([29b64e3](29b64e3)) ### Bug Fixes * bring the multi-agent coordinator online at boot ([#2007](#2007)) ([180b38a](180b38a)), closes [#1958](#1958) * full governance enforcement online ([#1957](#1957)) ([#2005](#2005)) ([4140fc5](4140fc5)) * harden anti-ghost-wiring gate and fix silently-dropped review agents ([#2000](#2000)) ([89b57ce](89b57ce)) * make coordination metrics live ([#1959](#1959)) ([#2012](#2012)) ([c4775e2](c4775e2)) * sandbox lifecycle dispatch (per-agent / per-task container reuse) ([#2008](#2008)) ([03d2587](03d2587)), closes [#1965](#1965) ### Documentation * add GitButler concept-only concurrency research ([#1978](#1978)) ([#2009](#2009)) ([9e4f5c1](9e4f5c1)) * honest-hybrid refresh of README, site, and design specs ([#2001](#2001)) ([f485bea](f485bea)) ### CI/CD * update apko lockfiles ([#2004](#2004)) ([e2b9eee](e2b9eee)) ### Maintenance * Update Infrastructure dependencies ([#2014](#2014)) ([0b16bdf](0b16bdf)) * Update Python dependencies ([#2015](#2015)) ([a7224bb](a7224bb)) * Update Web dependencies ([#2016](#2016)) ([7a7fe76](7a7fe76)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). --------- Co-authored-by: synthorg-repo-bot[bot] <279117679+synthorg-repo-bot[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
…2096) (#2097) ## Summary Audit follow-up to merged #1956 (PR #2003 "Agent runtime online + minimal safety spine"). The audit found 13 of 14 acceptance / scope items PASS; the only WEAK item (composed end-to-end test) was already covered by `tests/e2e/test_runtime_online_seam.py`. The single genuine gap was operator observability of the SecOps spine state at boot. This PR closes that gap. Closes #2096. ## What changed **Runtime-services boot log carries the safety-spine state** so operators reading `synthorg.log` can see, at a glance, whether the SecOps interceptor will be `active` / `shadow` / `disabled` once a provider runs. Two structured kwargs (`security_enabled`, `security_enforcement_mode`) added to all three `logger.info(API_APP_STARTUP, service="runtime_services", ...)` call sites in `src/synthorg/workers/runtime_builder.py` (the two no-provider branches in `_select_active_provider` and the provider-present branch in `build_runtime_services`). **Three pre-PR-review findings folded in by user request:** - `# module-kind: orchestrator` header added to `src/synthorg/workers/runtime_builder.py` (was defaulting to `code` tier with cap 500 but is an orchestrator at 840 gate-LOC; cap is now correctly `orchestrator` 600 with the baseline carrying the actual size). - `# module-kind: service` header added to `src/synthorg/workers/execution_service.py` (sibling classification gap; same shape). - Try / except around the `_build_runtime_coordinator` TaskGroup so a failure in decomposition / routing-scorer / workspace config resolve logs operator context (via `log_exception_redacted`) before propagating, instead of escaping silently. - New summary `logger.info(API_APP_STARTUP, mode="agent_engine_built", ...)` at the end of `build_runtime_services` reporting which optional subsystems (coordinator, work pipeline, red team, vision gate) wired, so operators can see the final boot shape in one line. ## Test plan - `tests/unit/workers/test_runtime_builder.py` adds a `TestBootLogSafetySpineState` class with three tests asserting the new `security_enabled` + `security_enforcement_mode` kwargs appear on each of the three boot branches (no-provider, empty-registry, provider-present), using `structlog.testing.capture_logs`. - Pre-push gates ran clean locally: ruff, ruff-format, mypy (affected), pytest-unit (affected), all convention gates including `module-size budget` and `no-ghost-wiring`. - E2E acceptance is already proven by `tests/e2e/test_runtime_online_seam.py::test_runtime_executes_task_through_seam_with_safety_spine` (unchanged). ## Review coverage Pre-reviewed by 4 agents per "reduced agent count" instruction: `code-reviewer` (APPROVED), `python-reviewer` (APPROVED), `logging-audit` (COMPLIANT), `conventions-enforcer` (1 MAJOR + 1 MEDIUM). Mandatory `docs-consistency` / `comment-quality-rot` / mini-pass agents skipped per explicit user instruction (no docs, no comment narrative, no ghost wiring introduced). All four valid findings (MAJOR + MEDIUM + 2 SUGGESTION) applied in this PR per user direction. ## Files - `src/synthorg/workers/runtime_builder.py` (+23 gate-LOC: 3 log sites + 2 local vars + try / except wrapper + summary log + module-kind header) - `src/synthorg/workers/execution_service.py` (module-kind header only; 0 gate-LOC) - `tests/unit/workers/test_runtime_builder.py` (new `TestBootLogSafetySpineState` class; +73 lines) - `scripts/_module_size_baseline.json` (mechanical baseline bump 809 to 840 for `runtime_builder.py`; user-approved) - `.codespellrc` (added `.test_durations.unit,.test_durations.integration` to skip list; was blocking pre-commit on a deliberate `parse_bool` test parametrize ID) ## Manifest No new `_ghost_wiring_manifest.txt` entries. `AgentEngine` is already `ENFORCED` from #1956.
1 participant
Summary
Wires the agent runtime online behind the provider-present switch (issue #1956, EPIC #1955 critical-path root).
synthorg.workers.runtime_builder.build_worker_execution_serviceconstructs a fully-wiredAgentEngine(LLM + sandboxed tools + per-call workspace + memory, governed by the SecOps safety spine) when a provider is configured, installed via the existingAppState.worker_execution_serviceseam;NoProviderExecutionServiceis the empty-company backstop.AgentRuntimeNotConfiguredError, code 4014); a provider added later wakes the runtime live via the existing setup-reinit path (no restart).ScriptedDrivertest/simulation provider + consolidated shared scripted test double.AgentEngineflippedPENDING -> ENFORCED;scripts/check_no_ghost_wiring.pypasses (Convention Rollout: lock ships with the fix).Pre-PR review
Reviewed by 20 review agents + 4 audit mini-pass agents before first push. 22 valid findings triaged and all implemented (
_audit/pre-pr-review/triage.md):_shared_app, coordination-wiring, and the coordination-controller fixture) rather than per-file.swap_worker_execution_servicenow holds_lazy_service_lock;swap_provider_registryserialised under a dedicated lock;SequencedResponseStrategy._indexguarded.resolve_agent_workspace_root_env); dead/lying docstring corrected.WORKERS_EXECUTION_SERVICE_TASK_NOT_FOUNDevent; ERROR log before boot-install failure propagates; INFO logs on lazy-service construction + bridge-config flip._next_statusbranch + unused constant removed; module/CLAUDE.md docs updated.ScriptedProvider()guard-double contract; MSW convention).Test plan
uv run ruff check src/ tests/+ruff format+uv run mypy src/ tests/: clean (3896 files).autonomy_resolver=None, post-run-missing, multi-provider warning, deep-workspace creation; e2e seam asserts the agent executed >=1 LLM turn.no-ghost-wiring).Closes #1956