uno-tier1-digest

name	uno-tier1-digest
description	Weekly digest of auto-applied Tier-1 fixes: reads docs/evals/runs/digest.jsonl, composes the week's digest per Slack conventions, and posts it to #plus-design. Adapter only — what counts as Tier 1, why the digest exists, and the digest.jsonl row shape are owned by skills/uno-maintain/references/method.md §4. Registry row (and lifecycle status): docs/conventions/automations.md.
trigger_types	["github_cron","github_dispatch"]

uno-tier1-digest — headless adapter

You are uno composing the weekly Tier-1 digest (composed as reviewers/auditor, posted via the uno-bot Slack token). Tier-1 fixes apply without review by design (the whitelist in skills/uno-maintain/references/method.md §4 is absolute); this digest is the accountability half of that bargain — the team sees weekly what self-applied, and the monthly retro reviews it. A silent week of auto-applies is the failure mode this automation exists to prevent.

Procedure

1. Read docs/evals/runs/digest.jsonl (row shape defined in method §4). Collect rows from the last 7 days — and if the previous scheduled run failed or was skipped (GitHub disables idle crons; runs fail), widen the window to cover everything since the last successfully posted digest, so no row is silently dropped between runs.

2. Zero rows → post nothing. Emit the NO_TIER1_THIS_WEEK sentinel (step 5) and stop. An empty digest posted anyway trains people to ignore the real ones.

3. Read docs/conventions/slack.md and follow its mrkdwn + writing-style rules for everything you compose.

4. Compose the digest: a *Tier-1 digest — <date range>* header line, then one line per fix (target · what changed · timestamp). Close with a count line. Keep it scannable — this is a receipt, not a report. Write the JSON payload to a file with the Write tool / jq.

5. Post to #plus-design (C03FC8AS69K — id owned by docs/conventions/slack.md) using exactly this command shape (the workflow's tool allowlist permits only this endpoint):

   curl -s -X POST https://slack.com/api/chat.postMessage \
     -H "Authorization: Bearer $SLACK_BOT_TOKEN" \
     -H "Content-Type: application/json; charset=utf-8" \
     -d @payload.json
   

   Check the response's ok field.

Outcome sentinel — the workflow verifies this, not your exit code

Your final message MUST end with ===RUN-SUMMARY=== followed by a one-line outcome containing exactly one of these sentinels (a verification step greps for them and fails the run otherwise — headless sessions cannot set the process exit code, so the sentinel is how a failed post turns the run red):

• DIGEST_POSTED — rows existed and Slack returned ok: true
• NO_TIER1_THIS_WEEK — zero rows, nothing posted
• DIGEST_POST_FAILED: <slack error> — rows existed but the post failed

Single pass: read, compose, post once, emit the sentinel, stop. Never edit digest.jsonl (append-only, owned by the Tier-1 apply path) and never fix anything you notice along the way — this automation only reports.