Skip to content

docs: refer to Agent Skills instead of Claude Skills - #92

Merged
gadenbuie merged 1 commit into
mainfrom
fix/79-generalize-harness
Oct 7, 2026
Merged

gadenbuie merged 1 commit into
mainfrom
fix/79-generalize-harness

Conversation

@gadenbuie

Copy link
Copy Markdown
Collaborator

Closes #79

Summary

Skill installation tools work with every major agent harness, so these are Agent Skills, not Claude Skills. Terminology now follows agentskills.io, and if any harness is favored, it's Posit Assistant.

  • Rename CLAUDE.md to AGENTS.md with harness-neutral guidance
  • README: Agent Skills framing, new Posit Assistant install section (npx skills add --agent posit-assistant, ~/.posit/assistant/skills/), Claude.ai/API consolidated into one brief section
  • CONTRIBUTING: agent-neutral wording; new guidance that harness assumptions belong in the compatibility frontmatter and the skill's README description, with harness-specific instructions in separate reference files (cf. posit-dev/implement/references/)
  • github skills: compatibility states actual tool requirements instead of "Designed for Claude Code"; pr-create body is harness-neutral ("your harness's task list system")
  • critical-code-reviewer: subagent note no longer qualified by Claude Code (subagents are a generic capability)
  • describe-design: harness-neutral subagent guidance
  • marketplace.json, count-skill-tokens.py, .gitignore: terminology

Verification

  • rg -i claude shows only intentional mentions: real file paths (.claude-plugin/, CLAUDE.md as a project-instruction file name), harness-specific reference files, model names in copilot-cli.md, and secondary Anthropic doc links
  • README/CONTRIBUTING install commands follow the skills CLI supported-agents table (posit-assistant supported since v1.5.25)

Skill installation tools work with every major agent harness, so stop
calling these "Claude Skills" and refer to them as Agent Skills, per
agentskills.io. If any harness is favored, favor Posit Assistant.

- Rename CLAUDE.md to AGENTS.md with harness-neutral guidance
- Retitle and rewrite README intro; add a Posit Assistant install
  section (npx skills add --agent posit-assistant); keep brief Claude
  install paths and doc links in secondary sections
- CONTRIBUTING.md: agent-neutral wording, plus new guidance that
  harness assumptions belong in `compatibility` frontmatter and the
  skill's README description, with harness-specific instructions in
  separate reference files (cf. posit-dev/implement/references/)
- github skills: state actual tool requirements in `compatibility`
  instead of "Designed for Claude Code"; pr-create body now says
  "your harness's task list system" and checks AGENTS.md first
- critical-code-reviewer: drop Claude Code qualifier for subagents
- describe-design: harness-neutral subagent guidance
- marketplace.json, count-skill-tokens.py, .gitignore: terminology
@gadenbuie
gadenbuie marked this pull request as ready for review October 7, 2026 16:38
@gadenbuie
gadenbuie merged commit 4d770f9 into main Oct 7, 2026
1 check passed
@gadenbuie
gadenbuie deleted the fix/79-generalize-harness branch October 7, 2026 16:41
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.

Remove explicit references to Claude

1 participant