prompt: route inference-method-cards-seed (#140) → library-dev (cloud) - #217
Merged
Merged
Conversation
Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018dV9h5h76hZ6trGZKWJFHe
Jammy2211
pushed a commit
that referenced
this pull request
Aug 24, 2026
Third instance of one family: a conductor deriving from prose while ignoring a declared header key. The ranker leg shipped as PyAutoBrain#217; the Bug Agent leg was recorded in that completion record and is still unfiled; this is the Intake Agent's sizing leg. analyse() hand-builds the prompt-shaped dict it hands the sizing faculty (_intake.py:268) and omits declared_header(text), which the file-loading path in _sizing.py:369 does fold in. estimate_difficulty therefore never sees a declared key, and _intake.py:320 writes the derived level into the header — silently replacing what the input asked for. Reproduced 2026-08-24: raw text declaring medium classified as too-large, unchanged when the full header block was prepended, so the header had to be hand-corrected after --apply. The prompt argues the reconciliation belongs in the sizing faculty both conductors already call, rather than per-conductor, and asks for the precedence rule to be decided explicitly. It also records that intake classified this bug report as a feature, from prose, while the difficulty leg was being written up — the work-type leg of the same family firing on its own bug report. The Feature Agent path is unaffected: it reports "declared medium but derived too-large — declared wins", which is #217 working as intended. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Hw4TUHAT7WjEXCJEvm3thw
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.
Files and routes the follow-on inference-programme docs task: seed the
results/notes/inference/methods/evidence cards (nautilus, nss, multi_start_prodigy, multi_start_adam) from the existing benchmark record, plus the recorded Phase 0(e) tick in PROGRAMME.md's phase table. Tracked as autolens_profiling#140; prompt inactive/inference_method_cards_seed.md; registered inactive.md(library-dev, cloud session, branchfeature/inference-method-cards-seed).🤖 Generated with Claude Code
https://claude.ai/code/session_018dV9h5h76hZ6trGZKWJFHe
Generated by Claude Code