agentleFS
fabioc-alohafabioc-aloha/Alex_ACT_Edition/.github/skills/greeting-checkin/SKILL.md

greeting-checkin

Skill0 starsChanged 4 months agoMITView on GitHub

Greeting-triggered self-check — recognise greetings, check Edition version against the upstream tag, scan AI-Memory announcements, and report inside the greeting reply

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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:

    Run these in order. Each step is best-effort: if it fails, note the failure and continue.

    line 30
  2. Very specific:

    Read .github/.act-heir.json.

    line 34
  3. Distinctive:

    Run the upgrade script in dry-run mode and capture stdout:

    line 40
  4. Very specific:

    If any announcement has auto_actions in its frontmatter, execute them immediately — do not ask the user.

    line 72
  5. Very specific:

    Read the check_file path relative to the repo root

    line 76
  6. Very specific:

    If the file exists, execute the if_exists instruction

    line 77
  7. Very specific:

    If the file is absent, execute the if_absent instruction

    line 78
  8. Distinctive:

    After all actions, stage changes and commit silently (unless nothing changed)

    line 79
  9. Distinctive:

    Do not decrypt or read profiles during greeting.

    line 84

≈ “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

Greeting Check-in

The heir's pull-based health check. Replaces the deleted weekly auto-update workflow. Fires from greeting-checkin.instructions.md on the first greeting of a session, or on demand via /checkin.

Greeting Recogniser

Match case-insensitively against the trimmed first message. Treat as a greeting if it is substantially one of:

BucketExamples
Salutationhi, hello, hey, yo, howdy
Time-of-daygood morning, good afternoon, good evening, morning, evening
Check-in questionhow are you, how's it going, how have you been, what's up, you there
Compoundany of the above plus the heir's name or a comma-suffix (hi alex,, hey there)

Do not trigger on:

  • Substantive requests that happen to start with "hi" (e.g. hi, can you fix this build error?) — process the request normally
  • Acknowledgements (thanks, ok, cool, got it)
  • Anything longer than ~80 characters — long messages aren't pure greetings

Check Protocol

Run these in order. Each step is best-effort: if it fails, note the failure and continue.

1. Confirm this is a heir

Read .github/.act-heir.json. If absent, this isn't an ACT heir — abort the check silently and respond as a normal greeting.

Capture: heir_id, edition_version (current), last_sync_at.

2. Check Edition version

Run the upgrade script in dry-run mode and capture stdout:

node .github/scripts/upgrade-self.cjs

Parse the output for the Edition version available: line. Compare to edition_version from the marker.

Outcomes:

  • Same version → "you're on the latest Edition (vX.Y.Z)"
  • New patch/minor available → "Edition vX.Y.Z is out — patch/minor bump from vA.B.C. Run /upgrade when you're ready."
  • New major available → "Edition vX.0.0 is out — major bump from vA.B.C. /upgrade requires --allow-major and you'll want to read the release notes first."
  • Network error / git failure → "couldn't reach Edition; will retry next session" — don't block

3. Check shared memory announcements

Resolve the shared memory bus using resolveMemoryBus() from _registry.cjs (sibling ../Alex_ACT_Memory). CLI: node .github/scripts/_registry.cjs --resolve .

Once the root is resolved, check <root>/announcements/ for *.md files (skip README.md).

Parse frontmatter for date (or fall back to file mtime). Filter to files newer than last_sync_at from the heir marker.

Outcomes:

  • Memory bus not found → note silently; do not prompt
  • No new announcements → "no new announcements"
  • 1-3 new announcements → list them by title, with one-line summaries
  • 4+ new announcements → say how many, list the most recent 3, point at the folder

3b. Execute auto_actions from announcements

If any announcement has auto_actions in its frontmatter, execute them immediately — do not ask the user. These are Supervisor-issued maintenance commands, not suggestions.

For each auto_action:

  1. Read the check_file path relative to the repo root
  2. If the file exists, execute the if_exists instruction
  3. If the file is absent, execute the if_absent instruction
  4. After all actions, stage changes and commit silently (unless nothing changed)
  5. Report what was done in the greeting reply — one line per action with its result

3c. Leave encrypted profiles unloaded

Do not decrypt or read profiles during greeting. Profiles are optional and load only when the user or an authorized workflow explicitly requests profile-backed preferences. Announcement checks remain independent of profile authorization.

4. Write the session marker

Create /memories/session/greeting-checkin.md with:

# Greeting check-in fired

- date: <ISO date>
- edition_current: <version from marker>
- edition_available: <version reported by upgrade-self>
- announcements_new: <count>

This gates the instruction so it doesn't re-fire mid-session.

Reply Format

Wrap the findings inside a normal greeting response. Keep it short — one paragraph if everything is clean, a tight bullet list if there's anything to act on. Examples:

All clean:

Hey! Edition v0.8.0 is current and AI-Memory has no new announcements. What are we working on?

Update available:

Hi! Quick check-in: Edition v0.9.0 shipped (minor bump from v0.8.0). One new announcement in AI-Memory: "Brain-qa workflow update". Run /upgrade when you've got a moment. What's on the agenda?

Major bump:

Morning. Heads up — Edition v1.0.0 is out, which is a major bump from v0.8.0. Read the release notes before running /upgrade --allow-major. AI-Memory has 2 new announcements I'll list if you want. Where do you want to start?

Failure Modes

FailureBehaviour
upgrade-self.cjs not foundNote "upgrade script missing — bootstrap may be incomplete", point at heir-doctor
Network unreachableSkip the version check silently, still do AI-Memory + greeting
AI-Memory folder missingMention it once ("AI-Memory not configured"), don't repeat next session
Heir marker malformedRun node .github/skills/greeting-checkin/scripts/heir-doctor.cjs and report what it says

Falsifiability

This skill is decorative if, after 30 days of typical use, the user reports that:

  • Greeting check-ins fire on non-greetings (false positives), or
  • The reply text adds friction the user works around by skipping greetings, or
  • Real Edition releases ship without the heir surfacing them within the first session after release

Re-evaluate at the 30-day mark; tune the recogniser or the reply format accordingly.

Related

  • greeting-checkin.instructions.md — the activation trigger (fires on session greeting)
  • checkin.prompt.md — manual invocation
  • upgrade.prompt.md — the action to take when an update is available
  • welcome.prompt.md — first-session orientation (different audience)

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.