Cross-Agent Compatibility¶
This marketplace keeps Claude Code support while adding documented registration paths for Codex and GitHub Copilot CLI.
Shared Instructions¶
Use AGENTS.md as the shared source of truth for repository instructions. Codex reads AGENTS.md before work and layers global, project, and nested instructions by directory precedence. GitHub Copilot supports repository agent instructions through AGENTS.md; Copilot cloud agent also supports root CLAUDE.md and GEMINI.md.
CLAUDE.md should remain a Claude Code adapter:
@AGENTS.md
## Claude Code Specific Instructions
Append Claude-only plugin, skill, command, or MCP guidance here.
General personal defaults belong at user scope, not copied into every
repository. The project plugin's install-user-instructions skill asks which
systems to configure, previews each destination, and manages only a marked
block while preserving surrounding content.
| System | Supported user instruction surface | Verification |
|---|---|---|
| Claude Code | ${CLAUDE_CONFIG_DIR:-~/.claude}/CLAUDE.md |
/memory |
| OpenAI Codex | ${CODEX_HOME:-~/.codex}/AGENTS.md |
Start a fresh Codex run and ask it to summarize current instructions |
| GitHub Copilot CLI | ${COPILOT_HOME:-~/.copilot}/copilot-instructions.md |
/instructions |
| Cursor | Cursor Settings > Rules (User Rules) | Inspect User Rules, then start a fresh Agent chat |
Cursor does not document a stable editable file path for UI User Rules, so the
installer provides a copy/paste block instead of fabricating a home-directory
file. Repository .cursor/rules remains project-scoped, while .cursorrules
is legacy.
Claude Code¶
Claude Code uses the existing marketplace manifest:
claude plugin marketplace add neuromechanist/research-skills
claude plugin install project@research-skills
The canonical Claude marketplace file remains .claude-plugin/marketplace.json.
Codex¶
Codex supports repository marketplaces at .agents/plugins/marketplace.json and legacy-compatible marketplaces at .claude-plugin/marketplace.json. This repo includes the native .agents/plugins/marketplace.json so a checkout can be used directly as a Codex marketplace root:
codex plugin marketplace add neuromechanist/research-skills
codex plugin marketplace add ./path/to/research-skills
Codex skills are directories containing SKILL.md files with name and description frontmatter. The native .codex-plugin/plugin.json manifests point at the existing plugins/<name>/skills/ trees. Claude-specific commands and agents stay declared in the .claude-plugin/plugin.json manifests; Copilot agent templates are exposed through native .github/plugin/plugin.json manifests.
Codex custom subagents are separate from plugin skills. Current Codex docs define custom agents as standalone TOML files under ~/.codex/agents/ for personal scope or .codex/agents/ for project scope, with required name, description, and developer_instructions fields. Optional fields include model, model_reasoning_effort, sandbox settings, MCP servers, and skills. /agent inspects and switches active threads. For review/QA surfaces in this marketplace, keep agents/templates/*.toml as opt-in Codex custom-agent templates.
The project workflow routes Codex roles by capability: Sol for architecture,
observation, supervision, and synthesis; Terra for bounded phase-plan
elaboration after architecture approval; and Luna for clear, repeatable
implementation, focused tests, routine review, and validation. Exact aliases
are used only when the current account exposes them. The project plugin ships
opt-in phase-planner.toml and implementation-worker.toml templates that pin
Terra and Luna for local Codex custom agents.
GitHub Copilot CLI¶
Copilot CLI supports marketplaces through .github/plugin/marketplace.json and also looks for .claude-plugin/marketplace.json. This repo includes .github/plugin/marketplace.json for the native Copilot path:
copilot plugin marketplace add neuromechanist/research-skills
copilot plugin marketplace browse research-skills
copilot plugin install project@research-skills
Copilot CLI plugin manifests support component path fields including skills, commands, and agents; agents points to directories containing .agent.md files. Every plugin in this repo now has a native per-plugin .github/plugin/plugin.json manifest. Plugins with review/QA agents expose agents/templates/ as the Copilot agents component path. The agents/*.md files remain Claude Code agent shells; agents/templates/*.agent.md are the Copilot custom-agent files.
Copilot custom-agent files use YAML frontmatter where description is required and the Markdown body defines the agent behavior. Optional frontmatter includes name, tools, model, target, disable-model-invocation, and user-invocable. Keep review/QA agent descriptions scoped to "invoked by the <skill> skill" so the skill remains the cross-agent trigger owner.
The native project plugin also exposes phase-planner and implementation-worker Copilot profiles. They use capability-class language instead of hard-coded model names because available Copilot model identifiers vary by environment.
Claude Code model routing¶
Claude Code uses Fable when available, otherwise Opus or best, for macro
design, observation, supervision, and synthesis. Sonnet implements approved,
detailed plans and performs routine focused review. Any unresolved architecture
or high-risk invariant escalates back to the lead model.
Agent lifecycle¶
All supported systems should close/remove completed one-off agents after their reports are incorporated. Retain an agent only for a named recurring role with a concrete next task. This avoids stale agent fleets and makes the active workflow observable.
GitHub Markdown bodies¶
Semantic line breaks remain the default for prose source. GitHub issue and pull-request bodies are the exception because physical newlines can affect rendering: keep each paragraph on one source line, separate paragraphs with blank lines, and do not insert sentence- or clause-level newlines inside a paragraph.
Sources¶
- OpenAI Codex AGENTS.md guide: https://developers.openai.com/codex/guides/agents-md
- OpenAI Codex plugin build guide: https://developers.openai.com/codex/plugins/build
- OpenAI Codex skills guide: https://developers.openai.com/codex/skills
- OpenAI Codex subagents guide: https://developers.openai.com/codex/subagents
- Anthropic Claude Code memory guide: https://code.claude.com/docs/en/memory
- Anthropic Claude Code subagents guide: https://code.claude.com/docs/en/sub-agents
- GitHub Copilot repository instructions: https://docs.github.com/en/copilot/how-tos/copilot-on-github/customize-copilot/add-custom-instructions/add-repository-instructions
- GitHub Copilot CLI user instructions: https://docs.github.com/en/copilot/how-tos/copilot-cli/customize-copilot/add-custom-instructions
- GitHub Copilot CLI config directory: https://docs.github.com/en/copilot/reference/copilot-cli-reference/cli-config-dir-reference
- GitHub Copilot custom instructions support: https://docs.github.com/en/copilot/reference/custom-instructions-support
- GitHub Copilot custom agents configuration: https://docs.github.com/en/copilot/reference/custom-agents-configuration
- GitHub Copilot CLI plugin reference: https://docs.github.com/en/copilot/reference/copilot-cli-reference/cli-plugin-reference
- Cursor rules: https://docs.cursor.com/context/rules