toasterson/akh-medu
Watch
1
0
Fork
You've already forked akh-medu
 0
Code Issues 12 Pull requests 41 Projects Releases 3 Packages 8 Wiki Activity Actions

P3-RL A2 — map deterministic engine replay to per-step rewards #365

Open
toasterson wants to merge 8 commits from claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay into main
pull from: claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
merge into: toasterson:main
Conversation 1 Commits 8 Files changed 3 +934 -8
toasterson commented 2026-09-12 07:31:37 +00:00
Owner
Copy link

Anima work item 01a08260-4aab-72b3-ac04-9f4009f1b111.

Bounded execution slice of WI-301, unpacked 2026-09-08 at the user's request. The parent preserves the original acceptance contract and source history.

Scope
Implement replay_trace -> reward vector with statement/step/token alignment using satisfied/step and broken_by_refs. Keep engine replay as the marking scheme; no LLM judge. Normalize executability credit per required hop (example.depth) and expose the landed redundancy signal for anti-farming.

Prerequisites
No unfinished hard prerequisite; existing main APIs are the starting point.

Acceptance and delivery
Known-good, known-bad and redundant-padding fixtures receive correctly attributed rewards; token alignment is explicit and stable. Extra executable recall lines cannot inflate normalized credit.

Execution notes
Re-ground against current main before changing code; the planning baseline was b000e52b58. Inspect the parent PR/branch and preserve useful existing work. Produce only this slice and its evidence. An implementation needs a durable Anima Change and exact-head review/CI; a findings item needs a linked findings Document. A local commit, draft proposal or unrun experiment is not proof that a prerequisite is available.

Anima work item `01a08260-4aab-72b3-ac04-9f4009f1b111`. Bounded execution slice of WI-301, unpacked 2026-09-08 at the user's request. The parent preserves the original acceptance contract and source history. Scope Implement replay_trace -> reward vector with statement/step/token alignment using satisfied/step and broken_by_refs. Keep engine replay as the marking scheme; no LLM judge. Normalize executability credit per required hop (example.depth) and expose the landed redundancy signal for anti-farming. Prerequisites No unfinished hard prerequisite; existing main APIs are the starting point. Acceptance and delivery Known-good, known-bad and redundant-padding fixtures receive correctly attributed rewards; token alignment is explicit and stable. Extra executable recall lines cannot inflate normalized credit. Execution notes Re-ground against current main before changing code; the planning baseline was b000e52b5884442bcc0f78d520803a7fb202de96. Inspect the parent PR/branch and preserve useful existing work. Produce only this slice and its evidence. An implementation needs a durable Anima Change and exact-head review/CI; a findings item needs a linked findings Document. A local commit, draft proposal or unrun experiment is not proof that a prerequisite is available.
P3-RL A2 — map deterministic engine replay to per-step rewards
Some checks failed
CI / check-seshat (pull_request) Failing after 15m3s
Details
CI / check-akh-medu (pull_request) Successful in 8m25s
Details
CI / publish-chart (pull_request) Successful in 2m28s
Details
CI / docker-seshd (pull_request) Successful in 15m19s
Details
CI / test-akh-medu (pull_request) Has been cancelled
Details
8814ff6326 
Anima implementation session.
P3-RL A2 — map deterministic engine replay to per-step rewards
All checks were successful
CI / check-seshat (pull_request) Successful in 2m47s
Details
CI / check-akh-medu (pull_request) Successful in 8m34s
Details
CI / publish-chart (pull_request) Successful in 1m47s
Details
CI / docker-seshd (pull_request) Successful in 6m53s
Details
CI / test-akh-medu (pull_request) Successful in 19m7s
Details
a21da4444e 
Anima implementation session.
toasterson changed title from WIP: P3-RL A2 — map deterministic engine replay to per-step rewards to P3-RL A2 — map deterministic engine replay to per-step rewards 2026-09-12 09:13:22 +00:00
P3-RL A2 — map deterministic engine replay to per-step rewards
Some checks failed
CI / check-seshat (pull_request) Successful in 2m52s
Details
CI / check-akh-medu (pull_request) Successful in 9m12s
Details
CI / docker-seshd (pull_request) Successful in 10m38s
Details
CI / publish-chart (pull_request) Failing after 18m5s
Details
CI / test-akh-medu (pull_request) Successful in 34m39s
Details
8dfed47f9f 
Anima implementation session.
toasterson force-pushed claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay from 8dfed47f9f
Some checks failed
CI / check-seshat (pull_request) Successful in 2m52s
Details
CI / check-akh-medu (pull_request) Successful in 9m12s
Details
CI / docker-seshd (pull_request) Successful in 10m38s
Details
CI / publish-chart (pull_request) Failing after 18m5s
Details
CI / test-akh-medu (pull_request) Successful in 34m39s
Details
to 9d20e37c6f
Some checks failed
CI / check-seshat (pull_request) Successful in 11m20s
Details
CI / check-akh-medu (pull_request) Successful in 21m24s
Details
CI / publish-chart (pull_request) Failing after 14m53s
Details
CI / docker-seshd (pull_request) Successful in 20m24s
Details
CI / test-akh-medu (pull_request) Failing after 1h19m38s
Details
2026-09-12 12:31:41 +00:00 Compare
Merge remote-tracking branch 'origin/main' into recovery/wi390-20260922
All checks were successful
CI / check-seshat (pull_request) Successful in 5m5s
Details
CI / test-sokar (pull_request) Successful in 7m7s
Details
CI / test-akh-medu (pull_request) Successful in 8m41s
Details
CI / docker-seshd (pull_request) Successful in 10m58s
Details
CI / test-session-platform (pull_request) Successful in 15m47s
Details
CI / check-akh-medu (pull_request) Successful in 13m9s
Details
CI / docker-sokard (pull_request) Successful in 13m21s
Details
CI / publish-chart (pull_request) Successful in 10m21s
Details
CI / docker-akh-transcripts (pull_request) Successful in 13m10s
Details
5928154cad
toasterson commented 2026-10-01 16:23:44 +00:00
Author
Owner
Copy link

Full-lane babysit 2026-10-01 — WI-390.

Exact head 5928154cadd9f982e358106dfddb50d1cdd4b267, current main ad441fe9e2359a1bf315eda28a4b072cc56cee2d. Real merge-tree clean; branch is not already on main.

Three-file replay reward implementation and good/bad/padding/alignment fixtures exist; merge-tree clean and hosted CI latest contexts all green. Native tracker result is old failed9d20e37c, not current5928154c; resync reports already in sync. Full current-head review/native CI gates still uncleared.

Individual disposition also recorded in the Anima WI thread. No approval or CI gate was overridden.

Full-lane babysit 2026-10-01 — WI-390. Exact head `5928154cadd9f982e358106dfddb50d1cdd4b267`, current main `ad441fe9e2359a1bf315eda28a4b072cc56cee2d`. Real merge-tree clean; branch is not already on main. Three-file replay reward implementation and good/bad/padding/alignment fixtures exist; merge-tree clean and hosted CI latest contexts all green. Native tracker result is old failed9d20e37c, not current5928154c; resync reports already in sync. Full current-head review/native CI gates still uncleared. Individual disposition also recorded in the Anima WI thread. No approval or CI gate was overridden.
fix(grug-data): align reward credit with valid engine replay
Some checks failed
CI / check-akh-medu (pull_request) Successful in 9m54s
Details
CI / docker-akh-transcripts (pull_request) Successful in 15m39s
Details
CI / test-session-platform (pull_request) Successful in 33m8s
Details
CI / docker-seshd (pull_request) Successful in 32m36s
Details
CI / test-sokar (pull_request) Successful in 53m20s
Details
CI / check-seshat (pull_request) Successful in 55m50s
Details
CI / publish-chart (pull_request) Successful in 1h42m27s
Details
CI / test-akh-medu (pull_request) Successful in 1h59m38s
Details
CI / docker-sokard (pull_request) Successful in 2h43m36s
Details
CI / platform-images-fleet (pull_request) Has been cancelled
Details
CI / platform-images-collect (pull_request) Has been cancelled
Details
CI / platform-images-core (pull_request) Successful in 20m13s
Details
CI / platform-images-home (pull_request) Successful in 34m21s
Details
CI / platform-images-daemons (pull_request) Successful in 1h17m56s
Details
5da210ac3d
Some checks failed
CI / check-akh-medu (pull_request) Successful in 9m54s
Details
CI / docker-akh-transcripts (pull_request) Successful in 15m39s
Details
CI / test-session-platform (pull_request) Successful in 33m8s
Details
CI / docker-seshd (pull_request) Successful in 32m36s
Details
CI / test-sokar (pull_request) Successful in 53m20s
Details
CI / check-seshat (pull_request) Successful in 55m50s
Details
CI / publish-chart (pull_request) Successful in 1h42m27s
Details
CI / test-akh-medu (pull_request) Successful in 1h59m38s
Details
CI / docker-sokard (pull_request) Successful in 2h43m36s
Details
CI / platform-images-fleet (pull_request) Has been cancelled
Details
CI / platform-images-collect (pull_request) Has been cancelled
Details
CI / platform-images-core (pull_request) Successful in 20m13s
Details
CI / platform-images-home (pull_request) Successful in 34m21s
Details
CI / platform-images-daemons (pull_request) Successful in 1h17m56s
Details
This pull request can be merged automatically.
This branch is out-of-date with the base branch
You are not authorized to merge this pull request.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay:claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay

Merge

Merge the changes and update on Forgejo.

Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.

git switch main
git merge --no-ff claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git rebase main
git switch main
git merge --ff-only claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git rebase main
git switch main
git merge --no-ff claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch main
git merge --squash claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch main
git merge --ff-only claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git switch main
git merge claude/wi-01a08260-p3-rl-a2-map-deterministic-engine-replay
git push origin main
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels   
follow-up
Documented operational follow-up

No labels
follow-up
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
toasterson/akh-medu!365
No description provided.