agentleFS
DScardini91DScardini91/agentic-os/.claude/skills/agentic-os-reference-sync/SKILL.md

agentic-os-reference-sync

Skill2 starsChanged 5 months agoMITView on GitHub

Orchestrates safe updates to the public agentic-os template from a private reference implementation. Use when syncing architecture, runtime contracts, README/Roadmap/Changelog, skills, agent patterns, hooks, or governance practices from a lived private OS into the sanitized public repo without leaking personal, client, or path-specific context. Triggers — "sync agentic-os from reference"; "update agentic-os based on the reference implementation"; "public template sync"; "sanitize reference changes for agentic-os"; "release agentic-os from reference"; "/agentic-os-reference-sync".

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 11 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    If the sync touches governance files, run a senior-advisor pressure-test before finalizing the recommendation.

    line 17
  2. Distinctive:

    Read only the bounded surfaces needed for the sync:

    line 35
  3. Distinctive:

    Do not copy private memory, client domains, personal identity, absolute paths, transcripts, or generated bundles.

    line 42
  4. Distinctive:

    Use this decision table:

    line 46
  5. Very specific:

    Before writing public docs, read references/sanitization-checklist.md.

    line 59
  6. Distinctive:

    Keep hooks and scripts dependency-light.

    line 63
  7. Distinctive:

    Avoid claiming parity across runtimes unless validated in that runtime.

    line 65
  8. Distinctive:

    If only one surface changes, document why the others do not need updates.

    line 79
  9. Distinctive:

    Run the public validation suite:

    line 83
  10. Distinctive:

    If a new or edited skill was added, also run:

    line 89
  11. Distinctive:

    If validation fails, fix before declaring the sync ready.

    line 95
  12. Distinctive:

    Do not push directly to main for a public repo unless the no-direct-merge ritual has been explicitly satisfied.

    line 107

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill: agentic-os-reference-sync

Mission

Convert advances from a private reference OS into a public, reusable agentic-os update. Preserve the architectural pattern, remove private implementation detail, and leave the public repo coherent across docs, skills, validation, and release notes.

Preconditions

  • You have read the public repo's root orientation and current README.md, ARCHITECTURE.md, CHANGELOG.md, and ROADMAP.md.
  • You know the private reference repo path or have been given a summarized delta.
  • The public repo is on a branch, not direct main, unless the principal explicitly authorizes a direct release.
  • If the sync touches governance files, run a senior-advisor pressure-test before finalizing the recommendation.

Workflow

1. Establish Current State

In the public repo:

git status --short --branch
git rev-list --left-right --count HEAD...origin/main
git log --oneline -8

Flag untracked archives, local commits ahead of origin, and dirty files before editing. Preserve unrelated work.

2. Inventory The Reference Delta

Read only the bounded surfaces needed for the sync:

  • Recent commits or release notes in the private reference.
  • Runtime contracts (AGENTS.md, CLAUDE.md, hook docs) when the delta concerns execution behavior.
  • Skills and agent patterns when the delta concerns reusable workflow.
  • Observability, telemetry, decision-log, and context-budget docs when the delta concerns governance.

Do not copy private memory, client domains, personal identity, absolute paths, transcripts, or generated bundles.

3. Classify Each Candidate

Use this decision table:

CandidatePublic action
Generic architecture principleAdd to ARCHITECTURE.md, control-plane/best-practices/, or agent-pattern docs
Reusable workflowAdd/update .claude/skills/<skill>/SKILL.md
Runtime-specific behaviorAdd to CLAUDE.md, AGENTS.md, or control-plane/AGENTS.md with runtime caveat
Validation or hook behaviorUpdate script/config only if it can run in a fresh fork without private dependencies
Private identity/client/path detailExclude; mention only as "reference implementation" if needed
Generated artifactRebuild from source or leave unchanged with rationale

4. Apply Sanitization

Before writing public docs, read references/sanitization-checklist.md. The minimum bar:

  • Replace names, clients, employers, absolute paths, account IDs, and private repo names with generic roles.
  • Convert "Kowalski does X" into "the reference implementation proved pattern X."
  • Keep hooks and scripts dependency-light; no private CLIs, local usernames, or one-off scheduler paths.
  • Distinguish implemented public behavior from reference-only behavior.
  • Avoid claiming parity across runtimes unless validated in that runtime.

5. Update Public Surfaces Together

For a material sync, keep these files coherent:

  • README.md — reader-facing pitch, quickstart, counts, status.
  • ARCHITECTURE.md — canonical architecture and design principles.
  • CHANGELOG.md — release entry with user-facing rationale.
  • ROADMAP.md — latest version and next bets.
  • control-plane/CLAUDE.md / AGENTS.md — runtime contracts, only when runtime behavior changes.
  • .claude/skills/ — orchestrating skills for repeated workflow.
  • control-plane/best-practices/README.md — index updates for new practices.

If only one surface changes, document why the others do not need updates.

6. Validate

Run the public validation suite:

bash scripts/validate-all.sh

If a new or edited skill was added, also run:

python3 /path/to/skill-creator/scripts/quick_validate.py .claude/skills/<skill-name>

If validation fails, fix before declaring the sync ready.

7. Release Staging

Prepare the branch for review:

git status --short
git diff --stat
git diff --check

Do not push directly to main for a public repo unless the no-direct-merge ritual has been explicitly satisfied. Prefer a PR with:

  • What advanced in the reference implementation.
  • What was deliberately excluded.
  • Validation results.
  • Any runtime caveats.

Output Format

Return:

## agentic-os reference sync

**Implemented:** <files / surfaces>
**Sanitized out:** <private/reference-only items excluded>
**Validation:** <commands + pass/fail>
**Release posture:** <branch, PR status, merge caveat>
**Open issues:** <only blockers or explicit follow-ups>

Anti-Patterns

  • Copying private memory or client-specific agents into the public template.
  • Updating ARCHITECTURE.md without aligning README status and changelog.
  • Claiming a runtime feature is shipped when only orientation docs were added.
  • Bumping release docs while validation is failing.
  • Direct-pushing to main on the public repo without the no-direct-merge ritual.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.