pipeline-digest
Skill84 starsChanged 7 days agoMITView on GitHub
Daily release-pipeline digest with a change gate — reports which plugins are pending release, whether main's CI is green, which branches went stale, and GitHub issue/PR activity, and notifies the operator only when that state actually moved since the last run. Use when the operator asks "what moved in the pipeline", "anything to ship", "pipeline digest", "any repo activity", or when the `pipeline-digest` routine fires. For a full on-demand pipeline snapshot regardless of change, use /release-status instead.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- gtapps/hermitdword for word, first seen Nov 7, 2026
Other versions1 project uses a version of this.
Word for word · 1
- gtapps/hermitd★ 84100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository29 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- bump-core-reqSkill
- claude-harness-canarySkill
- commitSkill
- docs-driftSkill
- downstream-impactSkill
- fleet-releaseSkill
- pipeline-digestSkill
- pre-release-reviewSkill
- pricing-refreshSkill
- releaseSkill
- release-statusSkill
- routine-test-pingSkill
- stale-proposalsSkill
- tackle-issueSkill
- test-runSkill
- bump-core-reqSkill
- commitSkill
- docs-driftSkill
- downstream-impactSkill
- fleet-releaseSkill
- pre-release-reviewSkill
- pricing-refreshSkill
- releaseSkill
- release-statusSkill
- routine-test-pingSkill
- stale-proposalsSkill
- test-runSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
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
- Distinctive: line 18
Run from the repo root. Two possible outputs:
- Very specific: line 41
Commit only when the digest reached the operator, or when there was nobody to send it to.
≈ “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
Pipeline Digest
Delta-gated wrapper over the pipeline facts /release-status reports in full. A quiet day costs one script call and sends nothing.
Steps
1. Collect
bun .claude/skills/pipeline-digest/scripts/digest.ts .hermitRun from the repo root. Two possible outputs:
NOCHANGE— the pipeline is exactly where it was last time. Say so in one line and stop. Do not send anything.CHANGED|<hash>followed by the digest body — continue.
2. Report and notify
Print the digest body in the conversation. Then send it to the operator:
.hermit/bin/hermitd-run channel-send .hermit --noticewith {"maintainer": "<the digest, channel-voiced>"} on stdin. The hermitd-run dispatcher resolves the installed plugin itself — do not try to derive a plugin root from this skill's Base directory, which points at .claude/, not the plugin.
Channel voice: plain language, no file paths, no slash commands, no cron strings. Plugin names and versions are fine — the maintainer chat is the technical audience. Lead with what the operator can act on.
3. Commit the hash
bun .claude/skills/pipeline-digest/scripts/digest.ts .hermit commit <hash>Commit only when the digest reached the operator, or when there was nobody to send it to. Read the JSON on stdout — the exit code alone is not enough, because the script exits 1 whenever delivered is false, no_channel included:
| Send result | Action |
"delivered":true (exit 0) | commit the hash |
"no_channel":true (exit 1), no channel enabled in config.json | commit the hash — there is no audience, so printing it was the delivery |
"no_channel":true (exit 1), but a channel is enabled | do not commit — the channel is configured and unreachable, which is a delivery failure, not an empty audience |
| exit 1 otherwise (undelivered or degraded) | do not commit, so the digest re-fires next run; report the failure per § Operator Notification |
| exit 2 (payload rejected) | do not commit; the payload is malformed — fix and retry |
A plain contributor checkout with no hermit state directory is fine: the digest prints, no_channel short-circuits the send, and commit is a silent no-op.
4. Reconcile the proposal queue
Independent of steps 1–3: runs whether the pipeline digest itself was CHANGED or NOCHANGE, since a merge can close a proposal without moving any pipeline fact.
LAST=$(cat .hermit/state/proposal-reconcile-sha.txt 2>/dev/null || echo "")
NOW=$(git rev-parse HEAD)
[ "$LAST" = "$NOW" ] && echo "SKIP|unchanged" || echo "RECONCILE|$NOW"On SKIP, stop here.
On RECONCILE|<sha>, run the collector:
bun .claude/skills/stale-proposals/scripts/collect-evidence.ts --status proposed,deferred,acceptedNONE|no-open-proposals — write <sha> to .hermit/state/proposal-reconcile-sha.txt and stop.
Otherwise, dispatch a general-purpose subagent at model: "sonnet" — the matching step needs real judgment (partial-delivery detection, same-day evidence ordering, substance matching across a renamed thing), not a checklist, and a wrong strong match silently closes live work. Give it stale-proposals/SKILL.md's Step 2 contract verbatim with the bundle path from the collector above, then follow that skill's Step 2b (completeness check) and Step 3 (auto-apply SHIPPED-STRONG matches) exactly as written there.
For any SHIPPED-WEAK / AGED verdict: never use AskUserQuestion here — this routine fires unattended, with nobody watching to answer it. Instead, for each one, queue a plain yes/no bounded ask:
bun plugins/hermitd/scripts/proposal.ts queue-micro .hermit <<'HERMIT_MP'
{"tier":1,"question":"<summarize the verdict's evidence and gap in one sentence>. Close it as resolved anyway?","proposal_id":"<PROP-ID>"}
HERMIT_MPso the operator can answer later from any channel-reachable turn — resolving one of these later patches the proposal to status=resolved (mirroring stale-proposals/SKILL.md Step 3) if the answer is yes, and leaves it untouched if no.
Send one channel notice summarizing what changed (auto-resolved count, queued-question count) — fold it into the same notice as the pipeline digest itself when both fired this run, or send it alone when the digest was NOCHANGE. Then write <sha> to .hermit/state/proposal-reconcile-sha.txt.
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.

