Skip to content

feat(agent): add behavioral instructions for non-Claude agents#6

Merged
xnoto merged 1 commit intomainfrom
feat/agent-behavioral-instructions
Mar 3, 2026
Merged

feat(agent): add behavioral instructions for non-Claude agents#6
xnoto merged 1 commit intomainfrom
feat/agent-behavioral-instructions

Conversation

@xnoto
Copy link
Owner

@xnoto xnoto commented Mar 3, 2026

Summary

  • Rewrote AGENTS.md into a comprehensive behavioral contract modeled on Claude Code's built-in discipline: investigation-first thinking, read-understand-plan-act-verify loop, anti-over-engineering, safety protocols, git workflow, and communication style
  • Added model-specific behavioral nudges to each agent config targeting known failure modes (GPT verbosity, Gemini path hallucination, DeepSeek/Kimi/GLM/MiniMax language drift, etc.)
  • Removed opus.md — Claude is used natively via Claude Code

Test plan

  • Verify AGENTS.md renders correctly on GitHub
  • Spot-check agent configs have valid frontmatter + body content
  • Run pre-commit hooks pass

🤖 Generated with Claude Code

Rewrite AGENTS.md from a sparse 29-line skeleton into a comprehensive
behavioral contract modeled on Claude Code's built-in discipline. Add
model-specific nudges to each agent config targeting known failure modes
(verbosity, hallucination, language drift, over-engineering). Remove
opus.md since Claude is used natively.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@xnoto xnoto merged commit 84be9c2 into main Mar 3, 2026
1 check passed
@xnoto xnoto deleted the feat/agent-behavioral-instructions branch March 3, 2026 08:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant