weekly-update

name	weekly-update
description	Top-level orchestration skill for reusable weekly updates. Synthesize recent PRs, rollouts, incidents, reviews, and blockers using packet-driven evidence collection and narrow read-only delegation, keep worker outputs proposal-grade, keep final classification and wording local, and update only a last-success marker after a reviewed plan clears apply gates.

Weekly Update

Use this skill to produce a reusable weekly update with packet-driven evidence collection, optional narrow read-only delegation, and local final adjudication.

Use When

• the user wants a weekly update grounded in repo state plus directly related GitHub evidence
• final classification, section placement, wording, and marker updates must stay local
• mini workers are used only for narrow packet analysis, not for the final writeup

Execution Roots

• <skill-dir>: directory containing this SKILL.md
• <python-bin>: concrete interpreter path; on Windows prefer a non-WindowsApps Python
• <runtime-root>: <repo-root>/.codex/tmp/packet-workflow/weekly-update/<run-id>/
• <packet-dir>: <runtime-root>/packets
• <eval-log-json>: <repo-root>/.codex/tmp/evaluation_logs/weekly-update/<run-id>.json

Entry

1. Run gh auth status; stop if authentication is missing.
2. Collect, lint, and build with <python-bin> -B <skill-dir>/scripts/collect_weekly_update_context.py, lint_weekly_update.py, and build_weekly_update_packets.py --result-output <build-result-json>.
3. Initialize the evaluation log, merge the build result, then read orchestrator.json, global_packet.json, mapping_packet.json, and only the focused packet needed for the current decision.
4. Draft weekly-update-plan.json locally, validate it with validate_weekly_update_plan.py, and run apply_weekly_update.py only from the reviewed plan.
5. Finalize the evaluation log after validate/apply results are recorded.

Continue Only If

• packet_worker_map remains the routing authority for delegated packets
• review-mode baselines, adjustments, worker recommendations, and token metrics stay in build/eval artifacts instead of widening runtime packets
• artifact_only candidates stay reference-only and do not surface as standalone weekly items
• the common path stays on global_packet.json, mapping_packet.json, and at most one focused packet unless an explicit reread reason is required
• apply reads only the reviewed plan fields overall_confidence, stop_reasons, and allow_marker_update

Stop When

• gh auth is missing or GitHub evidence required for the reporting window cannot be collected
• structured context is stale or the selected analysis ref cannot support the needed reread
• unresolved raw-reread candidates, low confidence, or marker-update gates block the plan
• incident-versus-blocker classification remains ambiguous after local review

Final Response

• state the exact reporting window
• say which packets and evidence categories drove the result and whether mini workers were used
• say whether the run stopped at planning, validation, or apply
• if blocked, name the blocker precisely

References

• references/weekly-update-contract.md
• references/architecture-note.md
• references/delegation-playbook.md
• references/weekly-update-evaluation-contract.md
• references/core-contract.md
• <python-bin> -B <skill-dir>/scripts/smoke_weekly_update.py --repo-root <repo-root>