Repository navigation
thinking-tools: self-rate — recommend an independent grader for subjective calls - #21
Open
mickzijdel wants to merge 1 commit into
Open
mickzijdel wants to merge 1 commit into
mickzijdel wants to merge 1 commit into
Conversation
…ctive calls A model grading its own subjective output (voice, tone, taste) inherits the same self-preferential bias a human author has toward their own draft. Add guidance to dispatch a separate subagent with the rubric for genuinely subjective/high-stakes grading, instead of always self-scoring. Source: commit-digest run against epoch-research/ai-productivity-digest feed (https://x.com/petergyang/status/2079257646939742542).
mickzijdel
pushed a commit
that referenced
this pull request
Jul 21, 2026
Scanned nateberkopec/dotfiles (5 new commits since Run 13's cutoff) and the epoch-research/ai-productivity-digest feed (20 new items). One suggestion cleared the bar and shipped as dev-hooks PR #21; the rest are logged with reasoning, including two explicit design-tradeoff conflicts (grill's one-question-at-a-time, worktree-setup's local-HEAD branching).
This was referenced Jul 21, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
commit-digestrun (feed:epoch-research/ai-productivity-digest, look-back 7 days) — this tip: for subjective outputs, use a separate verification agent with a rubric instead of self-grading, to avoid self-preferential bias.self-rateskill: for genuinely subjective or high-stakes grading (voice, tone, taste), dispatch a separate subagent as the grader (rubric + work only, no shared reasoning) instead of always self-scoring. References[[dispatching-parallel-agents]]for the dispatch mechanics.thinking-toolsplugin version 1.4.3 → 1.4.4 (patch, small addition to an existing skill).Why this one (of everything scanned)
The rest of this run's sources (7 non-merge commits from
nateberkopec/dotfiles, 37 feed items from the last 7 days) were personal-machine-specific (mise/pi/brew pins, a Heroku install tweak), already covered by an existing skill (api-scraping's HAR capture,premortem,weekly-automation-review,script-library,dispatching-parallel-agentsusage elsewhere), out of scope for this repo's domain (product-feature suggestions, Codex-specific features, unrelated topics), or directly conflicted with an existing skill's deliberate design (grill's one-question-at-a-time is intentional;worktree-setupbranches from local HEAD on purpose, not origin). Full breakdown logged toclaude/skipped-log. Nothing was Rails-specific, so no PR torails-toolkitthis run.Test plan
uv run pytest -q— 557 passedclaude plugin validate --strict .— passed🤖 Generated with Claude Code
Generated by Claude Code