chore: sync workflow templates

[stranske]

Sync Summary

Files Updated

• agents-weekly-metrics.yml: Weekly metrics - aggregates auto-pilot, keepalive, autofix and verifier metrics into summary reports
• aggregate_agent_metrics.py: Aggregates downloaded weekly agent metrics - required by agents-weekly-metrics.yml
• terminal_disposition.js: Machine-readable terminal disposition records and source summaries
• terminal_disposition_coverage.js: Warning-only terminal disposition source coverage preflight
• weekly_metrics_artifacts.js: Bounded weekly metrics artifact selection contract
• weekly_metrics_download_manifest.js: Weekly metrics artifact download and extraction manifest contract

Files Skipped

• pr-00-gate.yml: File exists and sync_mode is create_only
• ci.yml: File exists and sync_mode is create_only
• dependabot.yml: File exists and sync_mode is create_only
• llm_slots.json: None

---

Review Checklist

• CI passes with updated workflows
• No repo-specific customizations were overwritten

Source: stranske/Workflows
Source SHA: 9686fb727c01090a0a60095ef09bafdf10fb821b
Template hash: 8c201f72e3ed
Sync branch: sync/workflows-8c201f72e3ed
Consumer repo: stranske/Template
Manifest: .github/sync-manifest.yml

[stranske-keepalive]

⚠️ Action Required: Unable to determine source issue for PR #625. The PR title, branch name, or body must contain the issue number (e.g. #123, branch: issue-123, or the hidden marker ).

[Copilot]

Pull request overview

Syncs workflow template updates from stranske/Workflows to extend weekly metrics reporting (notably adding Codex CLI freshness metrics) and to harden/standardize metrics artifact download + terminal disposition normalization/coverage checks.

Changes:

• Add codex-cli-freshness as a first-class weekly metrics artifact family and summarize it in the aggregated report/contract.
• Centralize artifact name sanitization + enrich the weekly download manifest with more compact selection details and result-shape normalization.
• Refactor terminal disposition normalization of optional fields and adjust verifier model metadata coverage logic.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
scripts/aggregate_agent_metrics.py	Adds codex CLI freshness bucketing + summarization, and includes verifier CLI version aggregation.
.github/workflows/agents-weekly-metrics.yml	Uses shared safeArtifactPathSegment() helper when creating artifact directories during download.
.github/scripts/weekly_metrics_download_manifest.js	Adds compactSelectionDetails() and normalizes artifact download/unzip result shapes before updates/finalization.
.github/scripts/weekly_metrics_artifacts.js	Adds codex-cli-freshness to exact/prefixed artifact sets and priority families.
.github/scripts/terminal_disposition_coverage.js	Updates unsupported model env var fallback and changes “requires Codex model” logic.
.github/scripts/terminal_disposition.js	Introduces normalizeOptionalValue() to normalize numeric/boolean optional fields (incl. llm_cli_version).

[Copilot]

verifier_mode is normalized to an empty string when missing, but this condition now treats the empty value as "not evaluate". That changes behavior from “only enforce model metadata for explicit non-evaluate modes” to “enforce for missing/unknown modes too”, which can inflate missing-model-metadata counts and potentially trigger coverage warnings/hard-blocks for records that omit verifier_mode.

Consider restoring the prior guard (only enforce when verifier_mode is present and not evaluate) or explicitly defaulting missing verifier_mode to evaluate before this comparison.

Suggested change

	if verifier_mode != "evaluate":
	if verifier_mode and verifier_mode != "evaluate":

[Copilot]

In _summarise_codex_cli_freshness, status is counted using the raw string (no strip/lower), but later the summary derives outdated_records/latest_unavailable_records via exact-key lookups. Any casing/whitespace differences in incoming data will cause those derived counts to be wrong.

Normalize status (and potentially package/versions) consistently (e.g., strip().lower()) before incrementing counters and before doing key-based lookups.

Suggested change

	status = str(entry.get("status") or "unknown")
	statuses[status] += 1
	package = str(entry.get("package") or "unknown")
	packages[package] += 1
	pinned = str(entry.get("pinned_version") or "unknown")
	latest = str(entry.get("latest_version") or "unknown")
	status = str(entry.get("status") or "").strip().lower() or "unknown"
	statuses[status] += 1
	package = str(entry.get("package") or "").strip().lower() or "unknown"
	packages[package] += 1
	pinned = str(entry.get("pinned_version") or "").strip() or "unknown"
	latest = str(entry.get("latest_version") or "").strip() or "unknown"

[Copilot]

verifier_mode is cleaned to an empty string when absent, but requiresCodexModel now treats empty/unknown as requiring a Codex model ('' !== 'evaluate'). This is a behavior change from the previous logic and will start flagging missing model metadata for verifier records that omit verifier_mode.

If the intended behavior is “only require a Codex model when a non-evaluate mode is explicitly set”, reintroduce the Boolean(verifierMode) guard or default missing verifier_mode to evaluate before comparison.

Suggested change

	const requiresCodexModel = verifierMode !== 'evaluate';
	const requiresCodexModel = Boolean(verifierMode) && verifierMode !== 'evaluate';