error-scanlisted

# Error Scan Query Sentry for currently-unresolved issues across the org, diff them against a local ledger to find what's **new** or **regressed** since the last run, dynamically derive which checked-out repo each issue's `service` tag belongs to, and write a structured report. This skill makes **no code changes** — it reads and reports only. Use `/error-fix` to act on the findings. This skill hardcodes Sentry as the backend (no provider abstraction) — see `planning/error-patrol/PLAN.md` for the rationale. Support for other error-reporting backends is explicitly out of scope. --- ## Setup: Parse Arguments Before starting, check if any flags were passed: - `--summary` — print counts to stdout; skip writing `error-report.md` - `--dry-run` — run the full scan (Sentry API calls, service→repo derivation, diffing) but skip writing `error-report.md` **and** skip updating `state/error-patrol-ledger.json`. Print everything that would have been written to stdout instead. Use this to validate the skill end-to-end without mutating any state. --- ## Step 0: Preconditions 1. Confirm `SENTRY_ORG` and `SENTRY_AUTH_TOKEN` are set in the environment (`echo "org=$SENTRY_ORG token_set=$([ -n "$SENTRY_AUTH_TOKEN" ] && echo yes || echo no)"`). If either is unset or empty, print: ``` error-scan requires SENTRY_ORG and SENTRY_AUTH_TOKEN to be set in the environment. Skipping scan. ``` and stop. Do not write a report or touch the ledger. 2. Confirm `SHIPWRIGHT_REPO_DI