feat: inline imperative instructions in CLAUDE.md/AGENTS.md#190
Merged
Conversation
…tions CLAUDE.md and AGENTS.md now contain direct enforcement instructions instead of a passive skill router table. Based on Vercel eval data showing skills are skipped 56% of the time, and industry research on effective AGENTS.md patterns from 2,500+ repos. Key changes: - Always/When/Never three-tier boundary structure - RFC 2119 language (MUST, NEVER) for critical rules - Exact tool commands with parameters inline - Self-check checklist forcing model to verify its own work - ~77 lines, well within the <150 line adherence threshold Skills are still installed as bonus depth for Claude Code's skill system. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
abhigyanpatwari
added a commit
that referenced
this pull request
Mar 6, 2026
Regenerated CLAUDE.md and AGENTS.md using gitnexus@1.3.9 which replaces the old skill-router format with inline imperative instructions (PR #190). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
CrazyBunQnQ
pushed a commit
to CrazyBunQnQ/GitNexus
that referenced
this pull request
Mar 8, 2026
…tions (abhigyanpatwari#190) CLAUDE.md and AGENTS.md now contain direct enforcement instructions instead of a passive skill router table. Based on Vercel eval data showing skills are skipped 56% of the time, and industry research on effective AGENTS.md patterns from 2,500+ repos. Key changes: - Always/When/Never three-tier boundary structure - RFC 2119 language (MUST, NEVER) for critical rules - Exact tool commands with parameters inline - Self-check checklist forcing model to verify its own work - ~77 lines, well within the <150 line adherence threshold Skills are still installed as bonus depth for Claude Code's skill system. Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
CrazyBunQnQ
pushed a commit
to CrazyBunQnQ/GitNexus
that referenced
this pull request
Mar 8, 2026
Regenerated CLAUDE.md and AGENTS.md using gitnexus@1.3.9 which replaces the old skill-router format with inline imperative instructions (PR abhigyanpatwari#190). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
motolese
pushed a commit
to motolese/GitNexus
that referenced
this pull request
Apr 23, 2026
…tions (abhigyanpatwari#190) CLAUDE.md and AGENTS.md now contain direct enforcement instructions instead of a passive skill router table. Based on Vercel eval data showing skills are skipped 56% of the time, and industry research on effective AGENTS.md patterns from 2,500+ repos. Key changes: - Always/When/Never three-tier boundary structure - RFC 2119 language (MUST, NEVER) for critical rules - Exact tool commands with parameters inline - Self-check checklist forcing model to verify its own work - ~77 lines, well within the <150 line adherence threshold Skills are still installed as bonus depth for Claude Code's skill system. Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
motolese
pushed a commit
to motolese/GitNexus
that referenced
this pull request
Apr 23, 2026
Regenerated CLAUDE.md and AGENTS.md using gitnexus@1.3.9 which replaces the old skill-router format with inline imperative instructions (PR abhigyanpatwari#190). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
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.
Summary
Why
Vercel eval data shows skills are skipped 56% of the time — inline instructions achieve 100% pass rate. The previous router approach ("read this skill file") added an extra step that models frequently skipped. User feedback (Stephen/Neura project) confirmed models silently skip tool calls without mandatory rules.
What Changes for Users
After
npx gitnexus analyze, the generated CLAUDE.md/AGENTS.md will contain:Skills are still installed as bonus depth for Claude Code's skill system.
Test plan
ai-context.test.tsspecifically passes (marker upsert, no duplication)🤖 Generated with Claude Code