Adaptive Question Tree depth + Q-ID wording fix in Socratic recovery skill#525
Merged
rdmueller merged 1 commit intoMay 21, 2026
Conversation
…ry skill LLM-Coding#522 — Phase 1 decomposition is now adaptive, not near-fixed. Below the fixed second level a node is a leaf only when answerable from specific file:line evidence or definitively [OPEN]; coarse answers decompose further, capped at four levels below a fixed node. Tree depth tracks code density, so a large bounded context no longer produces one thin leaf per arc42 chapter. Directory-level evidence is rejected as a too-shallow signal. A post-prompt depth sanity-check and a corrected Q3.5 worked example demonstrate the pattern. LLM-Coding#524 — the Phase 2 prompt said "every claim must be traceable to a Q-ID", contradicting the schema (Q-IDs never appear in final documents). Reworded to match: claims trace back to a tree leaf as a build-time check; the output carries file:line evidence and (team answer) markers only. Touches the skill prompts/references, the socratic-code-theory-recovery contract, and the website skill page for consistency. Closes LLM-Coding#522, LLM-Coding#524 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Contributor
|
Caution Review failedPull request was closed or merged during review WalkthroughDer Pull Request präzisiert die Socratic Code-Theory Recovery-Methode: Phase-1-Decomposition wird adaptiv (Blätter entstehen nur mit konkreter ChangesAdaptive Question Tree mit Fixed Second Level
🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
2 tasks
This was referenced Jun 8, 2026
Merged
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
2 participants
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.
Summary
Two fixes to the Socratic Code-Theory Recovery skill, both surfaced by the Mistral Vibe tutorial walkthrough.
#522 — adaptive Question Tree depth. Phase 1 decomposed to a near-fixed depth regardless of bounded-context size, so a large context produced one thin leaf per arc42 chapter and Phase 2 could not synthesize a substantial chapter without inventing detail. Decomposition below the fixed second level is now adaptive: a node is a leaf only when answerable from specific
file:lineevidence or definitively[OPEN]; a coarse answer (a directory as evidence, one paragraph for a whole chapter) means the node decomposes further — capped at four levels below a fixed node as a runaway-recursion backstop. The fixed skeleton (Q1.1–Q5.5) is unchanged, so trees stay diffable; only depth varies.#524 — Q-ID wording contradiction.
phase-2-synthesize.mdopened with "Every claim must be traceable to a Q-ID", contradicting its own traceability rules andoutput-schema.md(Q-IDs never appear in final documents). Reworded: claims trace back to a tree leaf as a build-time check; output carriesfile:lineevidence and(team answer)markers only.Changes
prompts/phase-1-question-tree.md— adaptive decomposition rule, four-level cap, directory-evidence rejection, new post-prompt depth sanity-checkprompts/phase-2-synthesize.md— corrected the Q-ID wording (SKILL.md contradicts phase-2-synthesize.md on whether final docs carry Q-IDs #524)references/output-schema.md— adaptive-depth Q-ID scheme; directory is not valid evidencereferences/examples.md— theQ3.5worked example used directory-level evidence (the exact anti-pattern); rewritten as an adaptive parent node withfile:linechild leavesSKILL.md— adaptive-depth principle in the Phase 1 sectioncontracts.json—socratic-code-theory-recoverycontract aligned (EN + DE)docs/socratic-recovery-skill.adoc/.de.adoc— website skill page aligneddocs/changelog.adoc— 2026-05-21 skill-updates entryTest plan
contracts.jsonvalidates as JSONnode scripts/render-contracts.jsrenders without errorCloses #522
Closes #524
🤖 Generated with Claude Code
Summary by CodeRabbit
Dokumentation
file:lineoderfile::function-Zitate erlaubt, grobe Evidenzen erfordern weitere Zerlegung