skills/agent-config-audit/SKILL.md
Check whether Claude and Codex have equivalent access to shared agent resources, skills, hooks, plugins, MCP servers, permissions, startup behaviour, and provider-specific adapter config. Use when comparing agent environments, debugging missing capabilities after restart, or deciding whether to symlink a resource or configure a runtime.
npx skillsauth add sofer/.agents agent-config-auditInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Audit the hub-and-adapters setup for shared agent resources.
Use this rule of thumb:
Check only what is needed for the question, starting with:
~/.agents/AGENTS.md~/.agents/skills~/.agents/hooks~/.agents/docs/provider-compatibility.md~/.claude/settings.json~/.claude/settings.local.json~/.codex/config.toml~/.codex/config.json~/.claude/plugins or ~/.codex/pluginsUse rg, find, ls -la, sed, and tool help output before opening large logs or caches.
Report findings in these buckets:
shared and working: file-based resource is shared and adapter config points to it.shared but unverified: files or symlinks exist, but active-session behaviour has not been checked.Claude only: available through Claude config, hooks, plugins, MCP, or permissions only.Codex only: available through Codex config, hooks, plugins, MCP, or permissions only.missing adapter config: shared file exists but a runtime is not configured to use it.intentionally provider-specific: auth, history, cache, plugin cache, session state, or runtime-owned behaviour.unsafe to auto-fix: change could affect auth, secrets, external systems, permissions, production access, or destructive commands.For skills:
SKILL.md files under shared skill roots with the active session's available-skill list when available.For hooks:
For plugins:
~/.agents.For MCP and tools:
For permissions and safety:
~/.claude, ~/.codex, or global ~/.agents config unless the user explicitly asked for edits.~/.agents.Use this structure:
## Summary
<one paragraph>
## Findings
- `<bucket>`: <specific finding with file/config evidence>
## Recommended changes
- <change>, <why>, <whether approval is needed>
## Verification
- <what was checked>
- <what still needs a fresh session or manual confirmation>
testing
Record substantive skill use in an append-only local log. Use after choosing or invoking a non-system skill for real work, when a skill is inspected but not used, or when a skill fails to apply. Do not use for routine system skills or incidental file reads.
testing
Turn a vague or underspecified request into a self-contained problem statement. Use when the user has a rough idea, when a request would fail if handed directly to an agent, or before non-trivial work that needs shared understanding.
data-ai
Append a one-line learning to ~/.agents/learning-log.md. Use when the user types /learning, or when something genuinely worth remembering surfaced during work and the user confirms it should be captured.
testing
Define evaluation criteria, tests, or quality checks for a deliverable, implementation, workflow, or recurring agent task. Use when defining done, improving a test harness, or checking whether output meets its specification.