| name | flow-quick |
| description | Fast path — skip brainstorm/plan/waves for small tasks under 3 files. Estimate scope, dispatch a single specialist (welder by default), run verification, run a one-cycle critic gate, done. Escalates to flow:execute if scope or fixes exceed the limit. |
flow:quick — Fast Path (OpenClaw)
One agent, one pass, critic gate, done. For changes too small to justify the full pipeline.
Invocation: flow:quick "<task>" (inline task, no spec, no plan, no waves).
Steps
- Estimate scope. Read the likely file + its importers + test files. If ≥3 files, stop and
suggest
flow:brainstorm (offer --force to override). Under 3 files → continue.
- Detect language (
tsconfig.json/Cargo.toml/pyproject.toml).
- Dispatch one specialist via
sessions_spawn: welder by default (code/bug/param), guido for
design/naming, scout for research. Packet = task + scope files + language + "keep it minimal,
don't refactor unrelated code". If crew absent, do it directly with the same scope constraint.
- Verify: TS
npx tsc --noEmit && bun test; Py ruff check . && mypy . && pytest; Rust
cargo check && cargo test. Read full output + exit code.
- Critic gate:
sessions_spawn(agentId="fakoli-critic", …) on the modified files → PASS /
SHOULD FIX / MUST FIX.
- Evaluate:
- PASS → report files changed + verification + critic PASS. Done (no finish step unless asked).
- MUST FIX → ONE fix cycle (spawn welder with the exact findings, re-verify, re-critic). Still
MUST FIX → stop and escalate to
flow:execute.
- SHOULD FIX → log suggestions, proceed as PASS.
Quick mode has no spec, no plan, no sentinel — intentionally limited. If a "small" task turns out
to need a design decision, stop and route to flow:brainstorm.
