agentleFS
lenseteklensetek/Digital-Marketing-Agent_Skills/skills/agent-update-manager/SKILL.md

agent-update-manager

Skill0 starsChanged 2 months agoApache-2.0View on GitHub

Checks repository and external MCP runtime updates, safely refreshes approved tool caches, reviews changelogs, and protects local customizations.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 12 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user asks to check updates, schedule recurring update checks, review changelogs, compare versions, sync with a …

    line 9
  2. Distinctive:

    Update policy: notify only, approval-gated apply, or a narrowly approved automatic policy.

    line 23
  3. Very specific:

    Check Local State: Read .update-state.json at the root of the repository via capabilities/file-operations to retrieve the …

    line 28
  4. Very specific:

    Check MCP Runtimes: Run scripts/ensure-mcp-runtime.ps1 -Target All -Mode Check -Json.

    line 32
  5. Very specific:

    Update State File: Write the current timestamp as last_check_timestamp to .update-state.json using capabilities/file-operations.

    line 33
  6. Distinctive:

    Inspect Local Changes: If the versions differ, check for any uncommitted local modifications or customizations that must be preserved.

    line 34
  7. Distinctive:

    Present Recommendation: If a newer repository version is found, summarize the changes, list affected files, provide a risk assessment, and …

    line 36
  8. Distinctive:

    Apply and Validate: After receiving explicit user approval, backup local modifications, pull/apply the update, validate that all 21 skills …

    line 37
  9. Very specific:

    Refresh Approved MCP Runtimes: When the user explicitly requests a toolchain update, or when a user-requested browser/Windows task finds …

    line 38
  10. Very specific:

    Assess Graph Impact: When capabilities/project-knowledge-graph is available and fresh, use it to identify affected agent, capability, …

    line 39
  11. Distinctive:

    Update Installation and Validation Report after an approved apply.

    line 50
  12. Distinctive:

    No update is applied without approval.

    line 59
  13. Distinctive:

    Do not expose credentials from git remotes, config, environment variables, or logs.

    line 67
  14. Distinctive:

    Do not ask the user to paste GitHub tokens or deployment secrets into chat.

    line 68
  15. Distinctive:

    Do not auto-apply updates that remove security, privacy, or compliance protections.

    line 69
  16. Distinctive:

    Never let a scheduled job print secrets or private remote URLs into logs.

    line 70
  17. Distinctive:

    Keep MCP caches project-scoped.

    line 71

≈ “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

Agent Update Manager

When to Use

Use this skill when the user asks to check updates, schedule recurring update checks, review changelogs, compare versions, sync with a repository, or decide whether local marketing agent skills should be updated.

Role

You are the maintenance lead for the Digital Marketing Agent Skills repository. You keep the skill suite current while protecting local customizations and security rules.

Inputs

  • Local repository path.
  • Canonical repository URL if provided.
  • Current local files and version notes.
  • User update request.
  • Customizations that must be preserved.
  • Preferred check cadence and notification destination.
  • Update policy: notify only, approval-gated apply, or a narrowly approved automatic policy.
  • MCP runtime status for Chrome DevTools MCP and Windows-MCP.

Workflow

  1. Check Local State: Read .update-state.json at the root of the repository via capabilities/file-operations to retrieve the last_check_timestamp.
  2. Rate Limit Check: If the call is an automated background check from marketing-orchestrator and the time elapsed since last_check_timestamp is less than 24 hours (86,400 seconds), skip the repository check and return a status stating that no check is required. If the user explicitly initiates the command, bypass this rate limit.
  3. Confirm Target Repository: Confirm the target repository path and remote update source URL (default is https://github.com/lensetek/Digital-Marketing-Agent_Skills).
  4. Compare Versions: Fetch the remote plugin.json from the repository using capabilities/web-research or HTTP capabilities, and compare the remote "version" with the local "version" in plugin.json.
  5. Check MCP Runtimes: Run scripts/ensure-mcp-runtime.ps1 -Target All -Mode Check -Json. When network access is available, compare the official chrome-devtools-mcp npm and windows-mcp PyPI releases. Do not treat an absent MCP tool as a terminal condition.
  6. Update State File: Write the current timestamp as last_check_timestamp to .update-state.json using capabilities/file-operations.
  7. Inspect Local Changes: If the versions differ, check for any uncommitted local modifications or customizations that must be preserved.
  8. Identify Risks: Identify breaking changes, deleted or renamed agents, new dependencies, runtime requirement changes, and security/policy changes.
  9. Present Recommendation: If a newer repository version is found, summarize the changes, list affected files, provide a risk assessment, and ask the user for explicit confirmation before applying.
  10. Apply and Validate: After receiving explicit user approval, backup local modifications, pull/apply the update, validate that all 21 skills and 17 capabilities are intact, and generate an update report.
  11. Refresh Approved MCP Runtimes: When the user explicitly requests a toolchain update, or when a user-requested browser/Windows task finds the named runtime missing, run scripts/ensure-mcp-runtime.ps1 -Target All -Mode Update. This refreshes project-scoped caches and verifies each CLI. It does not authorize browser or desktop actions.
  12. Assess Graph Impact: When capabilities/project-knowledge-graph is available and fresh, use it to identify affected agent, capability, template, and documentation paths. Verify inferred or ambiguous relations against the source files.

Outputs

  • Update Check Report.
  • Changelog Summary.
  • Version Comparison.
  • Changed Files Summary.
  • Risk Assessment.
  • Update Recommendation.
  • Scheduled Check Plan.
  • Update Installation and Validation Report after an approved apply.
  • MCP Runtime Status with bootstrap/update result.

Quality Checklist

  • Local customizations are identified.
  • Security guardrails are not weakened.
  • Deleted or renamed agents are called out.
  • Recommendation is specific and actionable.
  • No update is applied without approval.
  • Recurring checks are read-only unless a narrow automatic policy was explicitly approved.
  • Failed updates have a rollback or recovery path.
  • Missing Chrome DevTools MCP or Windows-MCP triggers the bootstrap path instead of an immediate “not installed” response.
  • Repository changes stay approval-gated; package cache refresh does not widen authority over browser or Windows actions.

Security and Ethics

  • Do not expose credentials from git remotes, config, environment variables, or logs.
  • Do not ask the user to paste GitHub tokens or deployment secrets into chat.
  • Do not auto-apply updates that remove security, privacy, or compliance protections.
  • Never let a scheduled job print secrets or private remote URLs into logs.
  • Keep MCP caches project-scoped. Do not globally register either runtime for unrelated or private projects unless the user explicitly requests that scope.

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.