Skip to content

v0.9.3

Latest
Latest

Choose a tag to compare

View all tags
@hsnice16 hsnice16 released this 24 Oct 12:37
· 24 commits to main since this release
v0.9.3
This tag was signed with the committer’s verified signature.
hsnice16 Himanshu Singh
SSH Key Fingerprint: 9C+q1//ZT3no8M5n1U9Vtl5QKts2zfNUwU6f9sHLbQc
Verified
Learn about vigilant mode.
fee4c6f
This commit was signed with the committer’s verified signature.
hsnice16 Himanshu Singh
SSH Key Fingerprint: 9C+q1//ZT3no8M5n1U9Vtl5QKts2zfNUwU6f9sHLbQc
Verified
Learn about vigilant mode.

Every agent checked against its docs

  • A new check, agent skills: .agents/skills/<name>/SKILL.md, which Codex, Cursor, Devin, Gemini CLI, Kimi Code CLI, OpenHands and Pi load when a task matches.
  • Each agent's weights, rationale and read list re-checked against its current docs. GPT-5 Codex is now Codex and Kimi CLI is Kimi Code CLI; /agents shows that Cursor, Devin, OpenHands and Pi also read CLAUDE.md.
  • Cursor rules in subfolders, or in a subdirectory's own .cursor/rules/, now count, as do .claude/CLAUDE.md, .devin/blueprint.yaml and .cursor/environment.json. A GEMINI.md that imports AGENTS.md is measured with it.
  • The Aider check no longer passes on .aider.conf.yaml, which Aider never loads, and gives full credit only to a config that sets a test, lint or read option.

Full Changelog: v0.9.2...v0.9.3

Assets 2
Loading