feat(state): persist last_prompt_tokens / last_completion_tokens per call by OmarB97 · Pull Request #1 · OmarB97/hermes-agent

OmarB97 · 2026-05-25T14:52:30Z

Summary

Persist a per-call snapshot of the most-recent API call's prompt and
completion sizes on the sessions table. Cumulative input_tokens /
output_tokens mix every call together, so downstream consumers
(dashboards, monitoring tools) currently can't show "current
context-window size" without reconstructing it from per-call deltas
plus a cache model.

Motivation

MeshBoard's dashboard context chip was freezing on a static
system_prompt_size / 4 estimate for entire 30-minute dispatches
because state.db exposed no authoritative "current context size"
field. The dashboard shipped a workaround that uses system_prompt + session_input + session_output, but that double-counts when prompt
caching is active and misses tool-result re-introduction. The
authoritative fix is for Hermes to write the actual prompt_tokens
value the API returned for the most recent call.

Changes

hermes_state.py:

SCHEMA_SQL: two new columns on sessions:
- last_prompt_tokens INTEGER DEFAULT 0
- last_completion_tokens INTEGER DEFAULT 0
  The declarative _reconcile_columns() (line 514) will auto-ALTER TABLE ADD COLUMN on existing databases, no version bump needed.
update_token_counts() accepts new optional kwargs
last_prompt_tokens / last_completion_tokens. They are SET on
every update (never accumulated) in both absolute and increment
modes — they snapshot the latest call only, not lifetime totals.
Pass None (default) to leave the existing snapshot untouched
(e.g. cost-only updates).

run_agent.py:

The CLI commit path at line ~13237 now passes
last_prompt_tokens=prompt_tokens and
last_completion_tokens=completion_tokens (the raw values from the
canonical usage record) on every per-call token persistence.

Tests

4 new tests in tests/test_hermes_state.py:

test_last_prompt_tokens_is_set_not_incremented — multiple updates
set, not accumulate
test_last_prompt_tokens_omitted_keeps_previous_value — None
default leaves the snapshot untouched (the cost-only-update case)
test_last_prompt_tokens_in_absolute_mode — the gateway path also
sets, not accumulates
test_existing_db_reconciles_last_prompt_tokens_columns —
declarative reconciliation adds the columns on pre-existing DBs

Test run

$ python3 -m pytest tests/test_hermes_state.py tests/test_hermes_state_wal_fallback.py
============================= 231 passed in 13.64s =============================

(226 pre-existing + 5 new — the test that uses pytest fixture db
exercises the absolute-mode path and the reconciliation alongside
the new explicit tests.)

Compatibility

Existing databases: declarative reconciliation adds the columns
with DEFAULT 0 on the next Hermes startup. No migration script
needed.
Existing code: update_token_counts() signature is backward-
compatible. The new kwargs default to None.
Forward-compat with the dashboard: MeshBoard's
session_context_usage._extract_context_usage direct_keys check
already reads last_prompt_tokens from state.db rows
(line ~298). Once this PR ships, the chip starts showing
authoritative current-context size automatically — no dashboard
change required.

…call Cumulative input_tokens / output_tokens are the only token-related state.db columns today, which means downstream consumers (dashboards, monitoring) can't show "current context-window size" without reconstructing it from per-call deltas + a cache model. Add a per-call snapshot of the most-recent API call's prompt / completion sizes: - New SCHEMA_SQL columns `last_prompt_tokens` and `last_completion_tokens` on the sessions table. Both INTEGER DEFAULT 0; the declarative reconciliation in _reconcile_columns picks them up on existing databases via ALTER TABLE ADD COLUMN, no version bump needed. - update_token_counts() accepts optional last_prompt_tokens / last_completion_tokens kwargs. They are SET on every update (never accumulated) in both absolute and increment modes — they represent the latest call only, not lifetime totals. Pass None (default) to leave the existing snapshot untouched for cost-only updates. - run_agent.py wires `prompt_tokens` and `completion_tokens` from the canonical usage record through on every API call commit. Tests: - last_prompt_tokens is SET not accumulated across calls - passing None keeps the prior value (no zero-out on cost-only updates) - absolute mode also sets, not accumulates - the declarative reconciliation adds the columns on pre-existing DBs Motivated by: the MeshBoard dashboard's "context" chip was freezing on the system-prompt-size estimate for an entire 30-min dispatch because state.db exposed no authoritative "current context size" field. With this column the chip can render the real last-prompt size directly (via the existing direct_keys path in session_context_usage.py).

github-actions · 2026-05-25T14:53:19Z

🔎 Lint report: `feat/state-db-last-prompt-tokens` vs `origin/main`

ruff

Total: 0 on HEAD, 0 on base (➖ 0)

🆕 New issues: none

✅ Fixed issues: none

Unchanged: 0 pre-existing issues carried over.

ty (type checker)

Total: 8386 on HEAD, 8386 on base (➖ 0)

🆕 New issues: none

✅ Fixed issues: none

Unchanged: 4424 pre-existing issues carried over.

Diagnostics are surfaced as warnings — this check never fails the build.

Pre-filter fallback_providers when the agent starts up — entries whose provider has no credentials (no API key in env, no custom_providers definition, no base_url with env var set) are dropped silently from the chain and a startup diagnostic lists them with a hint naming the env var that would activate them. Before: runtime fallback loop discovered each failure one-by-one via 401/402/429 errors → wall of red logs, multiple seconds of latency, confusing UX (the #1 source of "why does my fallback keep 401ing"). After: chain is clean from the first failed API call; startup prints e.g. "Fallback chain: skipped 4 unconfigured: deepseek, minimax, zai, kimi-moonshot" with per-entry hints like "set DEEPSEEK_API_KEY in ~/.hermes/.env" or "`hermes auth add nous`".

…ch#37677) Anthropic enforces two independent ceilings per image: 1. 5 MB encoded byte size 2. 8000 px longest side Hermes only guarded #1. A tall screenshot (e.g. 1200x12000 at 0.06 MB) passes every byte check but fails the pixel check, returning a non-retryable HTTP 400 that permanently bricks the conversation thread. Fixes: - error_classifier: add 'image dimensions exceed' pattern to _IMAGE_TOO_LARGE_PATTERNS so the 400 is classified as image_too_large and triggers the shrink/retry path instead of falling through to non-retryable error. - conversation_compression: check pixel dimensions (via Pillow) even when byte size is under the 4 MB target. If max(dims) > 8000, force shrink. - vision_tools._resize_image_for_vision: add optional max_dimension param. When set, images exceeding the pixel cap are downscaled even if they're under the byte budget. The resize loop now checks both byte AND pixel limits before accepting a candidate. Closes NousResearch#37677

…bes + test-leak fix (NousResearch#40909) * fix(gateway,windows): reliability — supervisor task, JOB breakaway, status --deep Three coordinated fixes for the Windows gateway reliability story: 1. CREATE_BREAKAWAY_FROM_JOB on every detached spawn The 'hermes update' triggered from the Electron Desktop GUI ran inside Electron's job object. Without breakaway, the post-update gateway watcher spawned by update — already DETACHED_PROCESS — was still reaped when Electron's job tore down, so the gateway never came back after a GUI-initiated update. Adds CREATE_BREAKAWAY_FROM_JOB (0x01000000) to: - hermes_cli/_subprocess_compat.py::windows_detach_flags() — used by every helper that calls windows_detach_popen_kwargs(), including launch_detached_profile_gateway_restart() - The watcher subprocess's own respawn snippet in hermes_cli/gateway.py (inlined flags so the watcher's child respawn also breaks away) _spawn_detached() in gateway_windows.py already had the flag; this change brings the rest of the codebase to parity. 2. Per-minute supervisor Scheduled Task — Windows equivalent of systemd Restart=always Introduces hermes_cli/gateway_supervisor.py and registers it as a second Scheduled Task ('Hermes_Gateway_Supervisor', SC MINUTE /MO 1, LIMITED rights) alongside the existing ONLOGON task. Every minute, the supervisor uses the same gateway.status.get_running_pid() probe as 'hermes gateway status' and, if no gateway is alive, calls gateway_windows._spawn_detached() (which now includes BREAKAWAY) to bring one back. Covers every crash mode, not just 'machine rebooted': taskkill, OOM, GUI update SIGTERM, parent job teardown. Cheap — one pythonw startup per minute when down, one PID-existence check per minute when up. Wired into both the schtasks-success and Startup-folder-fallback install paths via _install_supervisor_best_effort(), and removed in uninstall(). Best-effort: a failing supervisor install logs a warning but doesn't roll back the primary install. 3. 'hermes gateway status --deep' shows per-probe PASS/FAIL Replaces the existing terse '--deep' output (which only printed paths) with an actual diagnostic table: [1] PID file present [2] Lock file held by a live process [3] get_running_pid() result [4] _pid_exists(pid) — OS-level liveness [5] gateway_state.json (state + age) [6] Last lifecycle event from gateway-exit-diag.log When the high-level summary disagrees with reality, the user can see exactly which signal is lying. Test-leak fix ------------- tests/hermes_cli/test_gateway_wsl.py::TestGatewayCommandWSLMessages monkey-patched is_linux/is_wsl/supports_systemd_services to simulate WSL but did NOT stub is_windows(). On a Windows host, the dispatcher in _gateway_command_inner takes the is_windows() branch BEFORE the WSL guidance branch, so the test invoked gateway_windows.install() for real. install() writes to %APPDATA%\...\Startup\Hermes_Gateway.cmd — the REAL user Startup folder, never sandboxed by tmp_path — pointing at the test's pytest-of-<user>/pytest-<N>/.../gateway-service/ wrapper. When pytest tore down the tmp_path, every subsequent Windows login flashed a cmd.exe window that failed to find the missing target. Stubs is_windows=False on all four affected tests: test_install_wsl_no_systemd test_start_wsl_no_systemd test_status_wsl_running_manual test_status_wsl_not_running Defense-in-depth: _build_startup_launcher() now prefixes the launcher with 'if not exist <target> exit /b 0', so any future stale Startup entry silently no-ops instead of flashing a console window. Status enhancements ------------------- - status() now reports supervisor task presence alongside the existing schtasks/Startup info, and nudges the user to reinstall if the supervisor isn't registered. - Deep mode dumps both the supervisor task name + script path. * fix(gateway,windows): drop the per-minute supervisor task — keep breakaway + deep probes Earlier in this branch we added a per-minute schtasks-based supervisor to respawn the gateway after crashes / GUI-update SIGTERMs. The implementation flashed a brief console window on every firing, which stole window focus. We tried several variants: - cmd.exe wrapper invoking pythonw -> flashes (cmd.exe is console-subsystem) - schtasks /TR pointing at pythonw -> flashes (uv venv launcher pythonw is actually subsystem=Console, not GUI; it respawns the real pythonw) - schtasks /TR pointing at base uv -> still flashes (Task Scheduler-side conhost preallocation; documented Windows quirk) - XML registration with <Hidden>true> -> still flashes (<Hidden> only hides the task in the Task Scheduler UI, not the spawned window) Researched what leading projects do: - Ollama: GUI-subsystem tray exe + Startup-folder shortcut. No supervisor. - Tailscale: real Windows Service via SCM. Session 0, no console possible. - Syncthing: --no-console flag inside the binary + Startup folder. - openclaw: VBS Run(..., 0, False) wrapper. Suppresses the *window* but Super User Q971162 confirms focus-steal still occurs in some cases. None of these use a per-minute polling scheduled task. The 'auto-restart on crash' responsibility belongs INSIDE the daemon (Tailscale's in-process recovery / Ollama's monitor+worker pair) OR is delegated to the Windows Service Control Manager — not Task Scheduler. So this commit drops the supervisor entirely. The CREATE_BREAKAWAY_FROM_JOB fix in _subprocess_compat.py (from commit c1e5fa4) survives — that is the *real* fix for problem #2 (GUI-update kills gateway): the post-update watcher in launch_detached_profile_gateway_restart() now breaks out of Electron's job object, so the gateway respawn watcher survives the GUI quit and successfully respawns the gateway. Surviving from c1e5fa4: * CREATE_BREAKAWAY_FROM_JOB in hermes_cli/_subprocess_compat.py (fixes #2) * Inlined breakaway flag in the watcher respawn snippet in gateway.py * hermes gateway status --deep PASS/FAIL probes (fixes #1 — visibility) * 'if not exist <target> exit /b 0' guard in _build_startup_launcher (fixes #3 — silent no-op for stale Startup entries) * tests/hermes_cli/test_gateway_wsl.py is_windows=False stubs (root cause of #3 — pytest WSL tests no longer leak Startup entries on Win hosts) Removed in this commit: * hermes_cli/gateway_supervisor.py (entire file) * Supervisor section in hermes_cli/gateway_windows.py (~180 lines): get_supervisor_task_name, get_supervisor_script_path, _build_supervisor_cmd_script, _write_supervisor_script, _install_supervisor_task, is_supervisor_task_registered, _install_supervisor_best_effort * _install_supervisor_best_effort() calls in install() (3 spots) * supervisor cleanup block in uninstall() * supervisor display lines in status() / status(deep=True) Future direction (out of scope for this PR): the right place for Windows 'Restart=always' semantics is a real Windows Service installed via pywin32's win32serviceutil.ServiceFramework — session-0 isolation, SCM auto-restart, no console window possible. That's a meaningful next-PR project, not a band-aid. Tests: 51 pass / 2 pre-existing failures in tests/hermes_cli/test_gateway_{windows,wsl}.py (the 2 failures are TestSupportsSystemdServicesWSL cases that fail on origin/main too — unrelated to this PR).

Phase 1 of the pluggable cron-scheduler refactor (Axis B — the trigger). No call-site changes; this phase only makes the abstraction exist + tested in isolation. Task 1.1: cron/scheduler_provider.py — the EXPERIMENTAL CronScheduler ABC. Required surface is name + start; is_available()/stop() carry safe defaults. is_available has a no-network invariant. Docstring marks it experimental until the Chronos provider (Phase 4) validates the shape. Task 1.2: InProcessCronScheduler wraps the historical 60s ticker loop, calling cron.scheduler.tick(sync=False) exactly as the raw ticker does. Uses stop_event.wait(interval) for responsive stop (both raw tickers already do). Tests: ABC-is-abstract, default-is_available, the InProcess loop drives tick and stops, stop() no-op, and test_abc_growth_stays_additive (the forward-compat guard: required abstractmethods must stay exactly {name, start}, so the three Phase-4 hooks land as NON-abstract additions). tick() internals in cron/scheduler.py are byte-unchanged (only new file added). Phase 0 characterization tests still green. Full tests/cron/: 445 passed.

* fix(windows): harden gateway scheduled task * fix(windows): launch gateway scheduled task via console-less wscript The Scheduled Task ran the gateway through cmd.exe, which allocates a console. During logon Windows broadcasts CTRL_CLOSE_EVENT to console process groups, reaping cmd.exe and the half-initialized gateway with STATUS_CONTROL_C_EXIT (0xC000013A) - which Task Scheduler treats as a user cancel, so RestartOnFailure never fires and the gateway vanishes on every reboot (issue NousResearch#45599 root cause #1). Add a console-less .vbs launcher (wscript.exe -> pythonw.exe, both GUI-subsystem) mirroring the gateway.cmd env + argv, and point the task action at it. The .cmd stays for the Startup-folder fallback and /Run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Jeff <jeffrobodie@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ture get_copilot_api_token now returns (api_token, base_url); the auth-remove suppression test still mocked it as a bare string, mis-unpacking into the credential-pool seed path and failing with 'No credential #1'.

…_id signature churn Two independent bugs evicted the cached gateway AIAgent on every turn, preventing the prompt cache from ever warming: 1. Model normalization mismatch: the post-run fallback-eviction check compared _agent.model (stripped in AIAgent.__init__) against the raw _resolve_gateway_model() config string. For vendor-prefixed config on native providers (e.g. 'deepseek/deepseek-v4-pro' vs 'deepseek-v4-pro') this was always unequal, so the agent was evicted after every successful run. Normalize _cfg_model the same way (skip aggregators). 2. Discord triggering message_id leaked into the cached system prompt via build_session_context_prompt()'s Discord IDs block. message_id changes every turn, so the agent-cache signature (computed from the ephemeral prompt) changed every Discord turn -> rebuild every message. The id is now injected per-turn into the user message (where per-turn content belongs and does not touch the cache signature); the cached IDs block carries a static pointer to it, preserving reply/react/pin via the discord tools. Adapted from NousResearch#28846. Bug #1 fix is the contributor's; bug #2 reworked to be non-destructive (keeps the triggering-id capability instead of deleting it). Redundant auto-reset eviction (already on main via NousResearch#9893/NousResearch#48031) and the wrong-premise reset_context_note plumbing from the original PR were dropped. Co-authored-by: Hermes Agent <hermes@nousresearch.com>

… fail on '(empty)' sentinel Two related bugs caused subagent delegation to silently return empty summaries with 0 tokens when the user configured delegation.provider=bedrock alongside delegation.base_url=https://bedrock-runtime.<region>.amazonaws.com. Root cause #1 — misrouting in _resolve_delegation_credentials(): The configured_base_url branch unconditionally forced provider='custom' and api_mode='chat_completions', only specializing for chatgpt.com, anthropic, and kimi hosts. Bedrock (and other native-SDK providers) fell through as 'custom' + chat_completions, which then POSTed OpenAI-shaped JSON at Bedrock's native API. Bedrock rejected the payload and returned nothing, which looked like an empty LLM response to the child agent. Fix: when provider is one of {bedrock, vertex, google, google-genai}, skip the base_url short-circuit and fall through to resolve_runtime_provider(), which knows how to construct the proper SDK client. base_url can still be forwarded through that path for regional overrides. Root cause #2 — '(empty)' sentinel accepted as success: After N retries of empty LLM responses, run_agent.py emits the literal string '(empty)' as final_response. _run_single_child then hit `elif summary:` — '(empty)' is truthy, so status became 'completed' and the parent surfaced a blank result with no error. Users saw api_calls=4, tokens=0, duration~0.4s, status=completed. Fix: treat final_response.strip() == '(empty)' as a failure so the parent surfaces it instead of silently accepting zero-content 'success'. Both paths were reproduced in a live Hermes TUI session on us-west-2 Bedrock (provider=bedrock, model=us.anthropic.claude-sonnet-4-6) and are covered by new tests in tests/tools/test_delegate.py.

…onnect ladder can't freeze silently (NousResearch#66377) The Telegram gateway could go silently deaf for hours: the reconnect ladder stalled mid-way (e.g. "attempt 4/10, reconnecting in 40s" then nothing) while the process stayed active(running), so Restart=always never fired. Root class: every recovery path — the ladder's re-entry (_schedule_polling_recovery), the pending-update probe (_probe_pending_updates), and PTB's error callback — gates new recovery on _polling_error_task.done(). If that single task wedges on any hung await, all recovery returns early forever and nothing retries. The heartbeat loop is a separate task, so make it an independent, cause-agnostic watchdog: if the same recovery task stays in-flight past _POLLING_ERROR_TASK_STUCK_TIMEOUT (300s — well beyond a healthy ladder attempt's bounded stop+drain+start+backoff), force a retryable-fatal so the background reconnector rebuilds the adapter instead of relying on the frozen ladder. This guarantees progress regardless of *where* the stall is (issue direction #1), tracked locally so no task-assignment site needs to change. Also salvages @koduri-mahesh-bhushan-chowdary's NousResearch#66492 (drain-await timeout), which closes the one concrete wedge vector documented in the incident (_drain_polling_connections' unbounded shutdown()/initialize() on a wedged CLOSE-WAIT pool). The watchdog covers the rest of the class. Co-authored-by: Koduri Mahesh Bhushan Chowdary <mkoduri73@gmail.com>

…reaming Two real render-cost wins found by inspection (no behavior change): 1. Sidebar re-rendered on every stream token. $sessionStates is republished on every message delta (tens/sec during a turn), and the derived ID computeds ($workingSessionIds, $attentionSessionIds, $backgroundRunningSessionIds) allocated a fresh array each time. nanostores notifies on !==, so the whole ChatSidebar + every mounted row re-rendered per token even when the working/ attention/background set was unchanged. Return the previous array reference when the contents match → nanostores skips the notify unless the set actually changes. Turns streaming from O(visible rows)/token into O(0) for the sidebar. 2. Tool rows normalized the FULL uncapped detail every render. `looksRedundant` (lowercase + whitespace-collapse over the entire read_file/terminal payload) ran twice in the ToolEntry render body, so every completed tool re-normalized its whole output on every stream tick of the running message. Memoize on the view fields so it recomputes only when the tool's content changes. Both are correctness-preserving (stable refs + memoization). The CI stream scenario drives $messages directly, not the publishSessionState path, so it won't reflect #1 — verified by inspection.

Blocking #1 — gateway-connecting-overlay.tsx reduced-motion regression: the top `if (reduce) setPhase('gone')` fired unconditionally on mount whenever reduce-motion was on, so every OS reduced-motion user lost the CONNECTING overlay during cold boot entirely (jumped to 'gone' before the gateway was even open). The intent was to skip the exit *choreography*, not to skip showing the overlay. Removed the unconditional top block and the redundant nested preview block; kept only the third branch (`gatewayState === 'open' && shownRef.current` → `reduce ? 'gone' : 'text-out'`) which correctly gates the short-circuit on connect. Also fixed `if(reduce)` missing-space, 6-space misindent, and the same 3-line comment pasted three times. Nit #1 — tsconfig excludes e2e, so specs were never typechecked in CI. Added tsconfig.e2e.json (extends base, includes e2e/ + playwright.config.ts, adds @playwright/test types) and wired it into the typecheck script. This surfaced three latent type errors that are fixed in the same commit: - fix-electron-tracing.ts: `app._context` and `electron._playwright` are private APIs — added `as any` on the access before the existing cast. - playwright.config.ts: `reducedMotion: 'reduce'` directly under `use:` is not a valid UseOptions property in playwright 1.58; it's a BrowserContextOption accessed via `contextOptions: { reducedMotion: 'reduce' }`. The old form was silently ignored at runtime, so reduced-motion emulation wasn't actually active — screenshots could catch overlays mid-fade (exactly what the comment warned about). Nit #2 — fix-electron-tracing.ts reaches into Playwright internals (_playwright, _allContexts, _context) with no public contract. Added a header comment calling out the `@playwright/test` exact pin (=1.58.2) so a future bump knows to re-verify the private symbols still exist. Nit #3 — main.ts TEST_WORKER_INDEX block had stray 6-space indentation. Verified: tsc -p . && tsconfig.electron && tsconfig.e2e → 0 errors; vitest boot-failure-overlay (3/3) + boot-failure-reauth (21/21) pass; npm run build clean; playwright e2e/boot-failure.spec.ts 2/2 pass.

…native extension) unicode61 indexes a CJK run as ONE token, so 2-char Korean terms (일본, 구글, 우리, ...) can never match it and the trigram tokenizer needs >=3 chars per term — any query containing a 1-2 char CJK token falls through to a LIKE full-table scan (measured 3-6.4s CPU per query on a 6.8GB production state.db; the #1 base cost behind a 12.4s session_search average on CJK workloads). This ships a ~250-line loadable FTS5 tokenizer (no deps) that wraps unicode61: maximal CJK runs inside its tokens are re-emitted as overlapping character bigrams (Lucene CJKAnalyzer semantics), everything else passes through unchanged. FTS5 phrase semantics turn consecutive sub-tokens into exact substring matching down to 2-char terms at index speed. Build: native/fts5_cjk/build.sh -> ~/.hermes/lib/libfts5_cjk.so (override: HERMES_FTS5_CJK_SO). Salvaged from PR NousResearch#65544; the schema integration lands separately on the v23 external-content layout.

@the3asic

…add same-pid self-reclaim guard Hardening on top of the salvaged dead-PID lease reclamation from PR NousResearch#65775 (@the3asic): - Probe via psutil.pid_exists (hard dependency; CONTRIBUTING.md critical rule #1) with the contributor's os.kill(pid, 0) POSIX probe retained only as a scaffold-phase fallback when psutil is missing. - Same-process holders (pid == os.getpid()) are never probed and never self-reclaimed — another thread's live lease is owned by the lease refresher/release path. - Any probe doubt (exceptions, permission errors) conservatively keeps the lease until normal TTL expiry; Windows stays TTL-only. - Tests: psutil-first dead-pid reclaim (probe call pinned), os.kill fallback path, probe-doubt keeps lease, same-pid no self-reclaim, legacy holder + Windows paths assert NO probe via either API.

…ch#67140) The background write guard decided ownership from `isinstance(usage_rec, dict)`, so a local skill with NO usage record passed. That successful write called bump_patch(), which created a `created_by: null` record — and the identical write was refused from then on. "Allowed exactly once, then never" is a race with our own bookkeeping, not a policy. Reproduced on main: patch #1 succeeds, patch #2 with the same arguments is refused. Option B from the issue. Option A (split `session_review` from `scheduled_curator` and let the session fork patch user-owned skills it consulted) would widen autonomous write permission onto skills the user owns with no user present to consent — wrong direction for a no-user-present actor. - skill_manager_tool: missing and explicit-null records now resolve IDENTICALLY, both fail closed. The refusal names the reason and points at `hermes curator adopt <name>`. - background_review: both review prompts told the reviewer to patch any skill consulted in the session and claimed pinned skills could be improved, while enforcement refused both. Prompts now list pinned, external, and user-owned skills as protected, and tell the reviewer to RECOMMEND adoption instead of attempting a write that will be refused. - skill_usage: document that `created_by` is a curator-management policy flag, not a provenance claim, and add `is_curator_managed()` so call sites read as the question they ask. Field name retained — it is on disk in every `.usage.json` and renaming would strand those records. - curator CLI: `hermes curator list-unmanaged` itemizes unmanaged skills with the reason each is unmanaged (completes the NousResearch#67139 spec). Foreground writes are untouched: a user-directed edit to a user-owned skill still works, including on pinned skills. Sibling tests: 9 failures in test_skill_manager_tool.py were fixtures that created record-less skills to exercise OTHER guards (consolidation-delete, read-before-write) and relied on ownership falling through. Fixed at the fixture, since the real curator only ever operates on managed sediment. One test asserted the old "manually authored" wording; rewritten to assert the behavior contract instead of the string. Validation: 274 targeted tests + all 7 background-review files (60 tests) pass. E2E on a temp HERMES_HOME (30 checks) covers the flip, foreground writes, adoption unblocking, pin semantics, prompt/enforcement parity, and the new verb. Each new test sabotage-verified: revert the fix, confirm it goes red. Fixes NousResearch#67140

OmarB97 merged commit 511c914 into main May 25, 2026
12 of 14 checks passed

This was referenced May 25, 2026

feat(state): persist pending_prompt_tokens for in-flight call display #2

Merged

sync: one-time catch-up of fork main against NousResearch upstream (1265 commits) #6

Merged

fix: pre-filter unconfigured fallback providers at startup #8

Merged

OmarB97 mentioned this pull request May 31, 2026

fix(agent): continue after recoverable tool loop guardrail #41

Merged

OmarB97 mentioned this pull request Jul 4, 2026

fix(agent): final_response contract — fork-pin the month-old verified fix + cover post-June drift #253

Merged

OmarB97 mentioned this pull request Aug 2, 2026

test(gateway): make the fallback-chain tests independent of host config #299

Merged

12 tasks

Navigation Menu

feat(state): persist last_prompt_tokens / last_completion_tokens per call - #1

feat(state): persist last_prompt_tokens / last_completion_tokens per call#1
OmarB97 merged 1 commit into
mainfrom
feat/state-db-last-prompt-tokens

OmarB97 commented May 25, 2026

Uh oh!

github-actions Bot commented May 25, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

OmarB97 commented May 25, 2026

Summary

Motivation

Changes

Tests

Test run

Compatibility

Uh oh!

github-actions Bot commented May 25, 2026

🔎 Lint report: feat/state-db-last-prompt-tokens vs origin/main

ruff

ty (type checker)

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

🔎 Lint report: `feat/state-db-last-prompt-tokens` vs `origin/main`