agent-tuneup
Skill2 starsChanged 9 days agoMITView on GitHub
Use when the user asks to audit, troubleshoot or change agent configuration, permissions, MCP connections or native skill discovery.
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- babysitterSkill
- challengeSkill
- choose-designSkill
- design-loopSkill
- focus-groupSkill
- github-writingSkill
- implementSkill
- raise-issueSkill
- raise-prSkill
- review-loopSkill
- reviewSkill
- shipSkill
- test-writerSkill
- thinkSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Follow relevant instruction roots, aliases, MCP settings and precedence.
- Distinctive: line 28
Never broaden access merely to suppress prompts.
- Distinctive: line 49
Use native tools. Keep private configuration out of reports and external prompts without sharing authority.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Inspect and align agent wiring
Workflow
- Inspect the affected host, versions, configuration files and working tree. Follow relevant instruction roots, aliases, MCP settings and precedence; expand the inventory only as needed to answer the request. Preserve unrelated work. For T3 startup or permission diagnosis, read the shared dispatch preflight.
Trace effective settings to their owning source or generator. Distinguish the user's intended policy, configured permissions and observed runtime behavior. Consult current official formats where needed; do not assume provider semantics or hand-edit generated copies. Unknown ownership blocks dependent changes.
The optional JSON scanner flags broad permission strings without printing values. Use only selected authorized files; pointers may still be private. It does not evaluate precedence or provider semantics. Flags prompt investigation, not automatic fixes. Observed calls do not prove approval. Do not scan credentials or transcripts by default.
- Make minimal authorized changes at the owning source and regenerate through its owner. Respect explicit permission choices and preserve deny rules, ownership guards, private overrides, vendor skills, credentials and active jobs. Never broaden access merely to suppress prompts. Before live cutover, back up the exact managed before-state host-locally with restricted access and a restore map; do not impose live-cutover procedures on ordinary source-only edits.
- Inspect generated results and run relevant schema/fixture checks plus required owning-repository checks and review before rollout. After authorized rollout, verify effective permissions and discovery in fresh sessions. Installed files and passing fixtures alone do not establish live compatibility.
Outputs
Scale the report to the task: result, affected paths, checks and unresolved gaps.
For audits, include evidence and failure scenarios; for live changes, include
rollback details. Mark unknown precedence, semantics and unverified runtime behavior
plainly. Keep source validation, rollout and fresh-session acceptance separate.
Delegates return complete plain-text reports to their coordinator. In T3 Code,
only the main-thread coordinator uses html_preview then html_render for
substantive audit findings and configuration comparisons; use readable text when
native HTML is unavailable or the result is brief.
Boundaries
Use native tools. Keep private configuration out of reports and external prompts without sharing authority. This skill grants no extra authority to edit, publish, broaden access or deploy.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

