fix(cognition): graded misses teach provisionally — belief hygiene - #2465
Merged
Conversation
…omePushParams The new #[ts(export)] provenance types + the parent_alloy_hashes push param — committing generated bindings to clear the drift guard (Rust is the source, TS is generated; commit both per the ts-rs law). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LoTjvf5j3Ez13g6k8mRkFo
… self-verdict (belief hygiene) The false-failure incident (2026-08-24): a broken env graded her correct code ok=False, and the lesson said flatly 'I did NOT solve it' — a conviction about her ability formed from an environment bug. A PASS is strong evidence; a MISS is WEAK, noisy evidence (broken env, a one-off, or a real gap — indistinguishable at grade time). So a miss now streams as a PROVISIONAL single attempt: 'this attempt did not pass yet... one attempt, not a settled verdict on my ability... worth retrying. If a later attempt passes, that supersedes this.' Both share the task.id recall key, so the dream reviewer can supersede the miss when a retake passes — exactly what happened this weekend (cancel-async 0→pass). A pass stays confident. Redaction still scrubs the answer key. Battery: a_miss_teaches + learn_mode_lesson + ratchets green (4 passed, EXIT=0). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LoTjvf5j3Ez13g6k8mRkFo
joelteply
enabled auto-merge (squash)
August 25, 2026 15:19
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A single miss is weak, noisy evidence (the weekend's broken-env false-failure proved it), so a failed exam lesson streams as a provisional, retryable, supersedable single attempt — never a hardened 'I am bad at X'. A later pass on the same task supersedes it via the shared recall key. Passes stay confident.
🤖 Generated with Claude Code
https://claude.ai/code/session_01LoTjvf5j3Ez13g6k8mRkFo