fix-from-diagnostics

name	fix-from-diagnostics
description	[Internal pipeline stage — run by /auto self-heal and /implement validation when lint/type error volume is high; invoke directly only as a power user.] Turn compiler/linter output into a sharded work queue; fix per package with optional adversarial review — no full monorepo suite mid-shard.
argument-hint	[--tool tsc|eslint|ruff|mypy] [--from-file path | capture from last command]

Fix-From-Diagnostics Skill

Turn typechecker/linter failures into a Bun-style work queue: parse once → cyclic-dependency pre-pass → shard by package → fix per shard (optionally dual-review large shards) → re-check diagnostics → only then run the full suite / smoke.

Use when:

• tsc / mypy / ruff / eslint reports many errors (rough rule: ≥ 15 distinct findings, or ≥ 3 packages), or
• /auto self-heal classifies lint_format / type_error and a single-file fix is clearly insufficient, or
• /implement Step 6 validation fails with a wall of type/lint noise.

Do not use for one-off annotation fixes (≤ ~5 errors in one file) — fix those inline.

---

Process

1. Capture diagnostics

Run the project's type/lint command(s) and capture full stdout+stderr to a file, e.g.:

# examples — use project-manifest / package scripts when present
npx tsc --noEmit 2>&1 | tee /tmp/harness-tsc.txt
npx eslint . -f stylish 2>&1 |  /tmp/harness-eslint.txt
ruff check . 2>&1 |  /tmp/harness-ruff.txt
mypy . 2>&1 |  /tmp/harness-mypy.txt