Skip to content

fix(ooda): operator Signal + machine-checkable check for the kgpacks-rs full-parity goal - #4328

Open
rysweet wants to merge 1 commit into
mainfrom
fix/kgpacks-rs-parity-goal-signal-and-done-gate-check
Open

fix(ooda): operator Signal + machine-checkable check for the kgpacks-rs full-parity goal#4328
rysweet wants to merge 1 commit into
mainfrom
fix/kgpacks-rs-parity-goal-signal-and-done-gate-check

Conversation

@rysweet

@rysweet rysweet commented Jul 18, 2026

Copy link
Copy Markdown
Owner

What & why

The standing goal advance-rysweet-agent-kgpacks-rs-to-full-parity-f29bb15c kept hard-parking as blocked cycle after cycle with why=UNCLEAR-CRITERIA ("no tracked PR/issue the done-gate can verify"). PR #4321 already rewrote the done-gate to a machine-checkable finish signal (tracking issue #4321 CLOSED == full parity), but two steps from the escalation_triage.md runbook were still outstanding:

  1. the done-gate had no single command a check could run, and
  2. no plain-English operator Signal message had been sent.

This PR completes the course-correction. Decision: rewrite-done-gate — and, honestly, the work is NOT delivered yet: KGP-Q4, KGP-T3, KGP-Q5 remain OPEN, so the goal is deliberately not marked complete.

Changes

Notes

  • LOCAL-ONLY docs/scripts; no product behaviour changed.
  • Follows the exact pattern of the prior blocked-goal course-corrections (coin-benchmark, session-checkpoint).
  • Pre-commit (fmt + clippy -D warnings) and pre-push gates pass.

Co-authored-by: Copilot 223556219+Copilot@users.noreply.github.com

…acks-rs full-parity goal

The standing goal advance-rysweet-agent-kgpacks-rs-to-full-parity-f29bb15c
kept hard-parking as blocked with why=UNCLEAR-CRITERIA ("no tracked PR/issue
the done-gate can verify"). PR #4321 already rewrote the done-gate to a
machine-checkable finish signal (tracking issue #4321 CLOSED ==
full parity), but two escalation_triage.md steps were still outstanding: the
done-gate had no single command a check could run, and no plain-English
operator Signal message had been sent.

This completes the course-correction (rewrite-done-gate; the work is honestly
NOT delivered yet — KGP-Q4, KGP-T3, KGP-Q5 remain OPEN):

- scripts/check-agent-kgpacks-rs-parity-done-gate.sh — one command the done-gate
  can run. Exits 0 (certified complete) when issue #4321 is CLOSED; otherwise
  exits non-zero after printing the exact remaining in-scope criteria as the
  concrete next step, so the goal is never merely "stuck". Verified locally:
  correctly reports OPEN + remaining KGP-Q4/KGP-Q5/KGP-T3, exit 1.
- docs/operations/agent-kgpacks-rs-parity-goal-signal-2026-07-18.md — the
  jargon-free operator/Signal notification (no raw markers), honestly stating
  the finish line is now auto-checkable but three items remain (linked from
  mkdocs nav).
- Specs/agent-kgpacks-rs-parity.md — additive "Machine-checkable finish signal"
  section binding the done-gate to issue #4321 and the check script.

LOCAL-ONLY docs/scripts; no product behaviour changed.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
rysweet added a commit that referenced this pull request Jul 18, 2026
…#4322)

Two pre-existing flaky lib tests intermittently failed the `pre-commit`
verify gate (which runs cargo test directly, not Python hooks), red-flagging
a cluster of otherwise merge-ready PRs (#4322, #4324, #4325, #4328). Both
flakes are on origin/main and are not PR-introduced.

- concurrent_dispatch_parallelizes_and_respects_cap: replace the brittle
  wall-clock ratio assertion (`parallel*2 <= serial`) — which jittered under
  CI scheduler contention on identical commits (#4328) — with a directional
  `parallel < serial` check. Concurrency is still proven structurally by
  run_count==N, peak_parallel>=2, and peak_serial<=1.

- meeting_turn_records_full_enriched_prompt_tokens_not_bare_objective:
  isolate the cost ledger via a new `#[cfg(test)]` thread-local
  `LedgerPathGuard` seam in cost_tracking::ledger_path() instead of mutating
  the process-global HOME. The old approach raced with any concurrent HOME
  mutator; the thread-local override cannot be observed by other threads, so
  the test drops the `cognitive_memory` serial key entirely. Both override
  branches (thread-local + SIMARD_COST_LEDGER) are strictly cfg(test)-gated,
  so the production ledger path is byte-for-byte unchanged.

Coupled observability fix: the silently-swallowed cost-write failures in the
meeting/PTY paths (base_type_copilot/mod.rs) were `eprintln!` — converted to
structured `tracing::warn!`, satisfying the no-print!/println! constraint.

Additive and test-scoped; no production behaviour change.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
rysweet added a commit that referenced this pull request Jul 18, 2026
…4189)

The OODA/overseer done-gate opened a fresh done-gate PR every cycle for a
goal whose engineer had exited leaving an open, unmerged PR — flooding the
repo (3x coin-benchmark #4326/#4329/#4332, 4x kgpacks-parity
#4324/#4328/#4330/#4333). Dispatch lacked any dedup keyed on durable goal
identity.

Make goal->engineer done-gate emission idempotent on stable goal identity:

- Durable primary guard: new `goal_pr_emissions` ledger (schema v1->v2,
  forward-only, additive) with `record_goal_pr_emission` upsert
  (ON CONFLICT(goal_key) DO UPDATE) and an indexed `find_open_goal_pr_emission`
  open-lookup. UNIQUE(repo, pr_number) turns a second goal claiming the same
  PR into a visible conflict, not a silent duplicate. Rows are never deleted —
  a completed PR transitions `state`, so an emission outlives its engineer.
- Advisory secondary seam: pure, total, no-panic dedup bricks
  (`goal_dedup_key`, `parse_goal_key_trailer`, `find_open_pr_for_goal`) plus a
  `GoalPrRef` DTO and a `PrGhClient::list_open_goal_prs` default so `gh`
  open-PR reconciliation can adopt a pre-existing PR by the `Simard-Goal-Key:`
  body trailer (primary) or `engineer/{key}-` branch convention (fallback).
  The goal key is a one-way, boundary-safe sha256 of goal identity, never the
  goal title, so distinct goals are unaffected.

Additive and non-breaking (default `list_open_goal_prs` returns empty; existing
fakes compile unchanged); structured tracing + OTel only, no stray print!, no
Bridge naming. Regression coverage: 6 ledger contract tests, 27 dedup unit
tests, and a 3-case integration test proving same-goal cycles emit exactly one
PR while distinct goals each dispatch. Docs: concept, how-to runbook, and
ledger API reference, all linked from the index and mkdocs nav.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant