Garry's Opinionated OpenClaw/Hermes Agent Brain
TypeScript
docs: CHANGELOG and TODOS for the /ship review fixes and follow-ups
garrytanbcd3d33887c3
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex d1179ef058..5b13efc6be 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -18,14 +18,14 @@ In gbrain-evals' Cat 31 re-run the full loop and a one-shot rewrite both scored### Itemized changes-- One-shot rewrite is the default for the CLI, `run_skillopt`, `--all`, queued jobs and the dream-cycle phase. A job queued before this release without `optimizer_mode` keeps the loop it was queued for, and `--resume` continues in the mode its checkpoint ran (a changed mode is refused). Receipts carry `optimizer_mode`.+- One-shot rewrite is the default for the CLI, `run_skillopt`, `--all`, queued jobs and the dream-cycle phase. A job queued before this release without `optimizer_mode` keeps the loop it was queued for, and `--resume` continues in the mode its checkpoint ran (a changed mode is refused); resuming a one-shot run whose rewrite gave no usable reply retries the rewrite. Receipts carry `optimizer_mode`.- The optimizer's success criteria come from training tasks only, and a rule argument or LLM rubric is quoted only when every training task shares it; a per-task answer is described ("a task-specific required text"), never quoted. Before, every selection and test answer was shown to the optimizer. `criteriaText` replaces the generated criteria.-- Held-out results never reach the optimizer: the held-out check runs once, on the final champion, before anything is written, and a refused champion is not filed in the rejected-edit buffer. Legacy `held_out_regression` rows on disk are skipped when the buffer loads.+- Held-out results never reach the optimizer. The held-out check runs once, on the final champion, before anything is written, and a refused champion is not filed in the rejected-edit buffer. Legacy `held_out_regression` rows on disk are skipped when the buffer loads.- Only edits that were applied are filed as rejected or recorded as accepted; edits the learning-rate budget clipped stay free to propose again.- The success-reflect prompt states the edit schema, so its edits are no longer all dropped (#7257; carried from #8324).-- Library surface for benchmark harnesses (`src/core/skillopt/index.ts`): explicit `tasks: {train, sel, test?}` lists instead of a benchmark file; a `rolloutHook` that receives each scoring site's whole batch (baseline, training forward pass, selection gate, held-out, test) and whose budget refusal or malformed reply aborts the run without writing; `scoreSkillPerTask`; and receipt counts (`rollouts` by phase, `candidates_scored_on_sel`, `optimizer_calls`, `cost_scope`).+- Library surface for benchmark harnesses (`src/core/skillopt/index.ts`): explicit `tasks: {train, sel, test?}` lists instead of a benchmark file; a `rolloutHook` that receives each scoring site's whole batch (baseline, training forward pass, selection gate, held-out, test) and whose budget refusal or malformed reply aborts the run without writing (the hook runs outside the run's budget, gets an abort signal at the runtime deadline, and a batch still running or answered after it aborts the run); `scoreSkillPerTask`; and receipt counts (`rollouts` by phase, `candidates_scored_on_sel`, `optimizer_calls`, `cost_scope`).- A skill can add the reviewed read tools `recall` and `get_timeline` to its rollouts by naming them in its frontmatter `tools:`; rollouts never offer `add_timeline_entry` or any other write.-- Tests: `test/e2e/skillopt-wave-e1b1.serial.test.ts` (15 cases, each red on master), `test/skillopt/benchmark-hooks.test.ts`, and a mode case in `test/skillopt/models-four-paths.serial.test.ts`.+- Tests: `test/e2e/skillopt-wave-e1b1.serial.test.ts` (17 cases, each red before its fix), `test/skillopt/benchmark-hooks.test.ts`, and a mode case in `test/skillopt/models-four-paths.serial.test.ts`.## [0.60.541.0] - 2026-11-01diff --git a/TODOS.md b/TODOS.mdindex cdd12b7e40..f77ee573b9 100644--- a/TODOS.md+++ b/TODOS.md@@ -1,5 +1,9 @@# TODOS+## SkillOpt E1/B1 follow-ups (filed 2026-11-01, GBRA-274, #8408)++- [ ] **P2 — SkillOpt review follow-ups (#8415).** **What:** held-out deferral on abort and the receipt's pending-champion score; final test eval on an unchanged skill; preflight prices the test eval at 2 runs per task instead of 6; resume does not pin the hook, criteria or cost scope; `--resume` with an unloadable checkpoint starts fresh; the one-shot output-cap check runs after baseline spend; one accepted one-shot step can replace the whole body; cat31 config C changed meaning; stale docs and comments. **Fix:** per item in #8415. **Effort:** S-M each. **Priority:** P2.+## Write-cap wedge follow-ups (filed 2026-11-01, GBRA-271, #8368)- [ ] **P3 — `owner_marker_missing` can name the plain self-transfer when another process sees a drifted APFS birth time (#8381).** **What:** the refusal compares the reservation's birth time exactly in the refusing process, and the self-transfer compares it exactly in the CLI process; on APFS the two can read tens of ns apart (#6501), so a long-lived process that created the checkout and recorded its reservation can recommend a self-transfer the CLI then refuses `recovery_required`. Nothing is written either way. **Fix:** one identity rule for both sides. **Effort:** S. **Priority:** P3.