batch-port

name	batch-port
description	Dispatch the next ready porting candidate from the catalog. Session-resilient — dispatches ONE port per invocation. Re-run to continue the batch.
user_invocable	true

/batch-port — Autonomous Batch Porting

Dispatch the next best porting candidate from the catalog.

Usage

/batch-port                                # Next best Cat 1 candidate for a1200_accel
/batch-port cli a1200_accel                # Explicit category + profile
/batch-port cli stock_a1200                # Different hardware profile
/batch-port scripting a1200_accel          # Category 2 candidates

Session-Resilient Design

This skill dispatches ONE port per invocation. After the port completes (or the session ends), re-run /batch-port to dispatch the next candidate. The catalog.json status field IS the queue — it persists across sessions.

Algorithm

1. Run python3 scripts/catalog-score.py --score to re-score all candidates (accounts for any shim changes since last run)
2. Run python3 scripts/catalog-score.py --next 1 --category <cat> --profile <prof> to get the best candidate
3. If no ready candidates: print recommendations (top shim unlocks) and stop
4. Check publishable field — skip if false
5. Set candidate status: "in_progress" in catalog.json (atomic write)
6. Dispatch the catalog-engineer agent with the candidate details:
   • Agent dispatches /port-project for the candidate
   • For Cat 1: use --unattended mode (auto-answer decisions)
   • For Cat 2+: standard interactive mode
7. On success:
   • Move candidate data to ported[] in catalog.json
   • Collect measured binary size + stack from compiled port
   • Update PORTS.md with new row
   • Re-run scoring to update the catalog
8. On failure:
   • Set candidate status: "failed" with error reason
   • Continue (user re-runs for next candidate)

Default Arguments

• Category: cli (Cat 1 — the only fully autonomous category)
• Profile: a1200_accel (the standard FS-UAE test config)

Stale In-Progress

If a candidate has status: "in_progress" (stale from a crashed session), treat it as a retry candidate. Re-dispatch it before selecting a new one.

After Batch Completion

Run python3 scripts/catalog-score.py --status to see the updated dashboard. Run python3 scripts/catalog-score.py --diff data/catalog-before.json to see what changed (if you saved a snapshot before the batch).