agent-team-standards
Skill1 starsChanged 23 days agoMITView on GitHub
Lock a project's agent skill names and versions in a manifest, then report drift against that manifest. Use when a team needs reproducible skill versions across CI, cron, and developer machines.
In this repository46 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .codex/AGENTS.mdAGENTS.md
- .gemini/AGENTS.mdAGENTS.md
- .cursor/rules/agent-foundry-orchestrator.mdcCursor rule
- .cursor/rules/agent-foundry-skills.mdcCursor rule
- web/llms.txtllms.txt
- anti-slopSkill
- api-designSkill
- automation-pickSkill
- bottleneck-gatingSkill
- budget-guardSkill
- cache-identitySkill
- constraint-then-solveSkill
- context-optimizationSkill
- cron-troubleshootSkill
- e2e-test-strategySkill
- engram-routingSkill
- execution-observabilitySkill
- fallback-model-routingSkill
- feedback-loopSkill
- judge-integritySkill
- knowledge-extractSkill
- landscape-firstSkill
- measure-firstSkill
- plan-before-codeSkill
- plan-then-actSkill
- prompt-disciplineSkill
- pushback-when-wrongSkill
- quality-protocolSkill
- read-before-buildSkill
- re-verify-findingsSkill
- session-closeoutSkill
- session-distillSkill
- show-your-workSkill
- structured-output-contractSkill
- verify-firstSkill
- workflow-decomposeSkill
- agent-memory-vaultSkill
- autonomous-briefingSkill
- book-to-skillSkill
- chrome-devtools-mcp-bridgeSkill
- design-languageSkill
- funnel-pr-guardSkill
- persistent-memorySkill
- sql-migration-trioSkill
- token-compressionSkill
The file
Files
Agent Team Standards
A manifest records which skills and versions a project expects. The shipped script compares a live skills tree with that manifest. It does not install, upgrade, or remove anything.
Procedure
What ships
skills/optional/agent-team-standards/scripts/ats.py (standard library only).
The script scans HERMES_SKILLS_ROOT, which defaults to
~/AppData/Local/hermes/skills on Windows. That root must contain category
folders, each holding skill folders with a SKILL.md:
<HERMES_SKILLS_ROOT>/<category>/<skill>/SKILL.mdCommands
python skills/optional/agent-team-standards/scripts/ats.py lock --project-dir=. --output=agent-teams.lock.json
python skills/optional/agent-team-standards/scripts/ats.py validate --manifest=agent-teams.lock.json
python skills/optional/agent-team-standards/scripts/ats.py diff --manifest=agent-teams.lock.json
python skills/optional/agent-team-standards/scripts/ats.py apply --manifest=agent-teams.lock.jsonWhat each command actually does
| Command | Behavior |
lock | Scans the skills root and writes name, category, version, path, and installed for each skill. |
validate | Checks manifest shape, required fields, and duplicate names. |
diff | Lists matching, mismatched, extra, and missing skills. |
apply | Reports missing and out-of-date skills. It is a dry run and changes nothing. |
The config_fragments key is emitted empty. Do not expect it to carry models
or AGENTS.md content.
Manifest example
{
"schema_version": "1.0",
"project": "my-project",
"generated_at": "2026-08-13T00:00:00Z",
"skills": [
{
"name": "agent-memory-vault",
"category": "optional",
"version": "1.0.0",
"installed": true,
"path": "optional/agent-memory-vault"
}
],
"config_fragments": {}
}Installation is separate
To make a machine match the manifest, use your package manager or install
script, then run diff again. A green apply means "no drift reported", not
"installed".
Anti-patterns
- Pinning versions without recording where each skill came from.
- Editing the manifest by hand instead of regenerating it with
lock. - Treating
applyas an installer. - Locking every project to one global standard.
Verification Checklist
lockwrites a manifest whose skill count matches the scanned root.validateexits 0 on the generated manifest and nonzero on a broken one.diffnames every intentionally missing skill.applyoutput is understood as a report, not a change.
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.

