Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Replace [FLAGS] with desired combination from the table above.
Tell the user the log path so they can monitor with tail -f or check after completion.
The background worker also writes a durable run log under the run directory:
~/.local/ratchet/runs/{project_id}/{run_id}/worker.log.
Generation Mode
By default, the local runtime uses deterministic local embeddings and structured
heuristics. To delegate generation to the current host agent CLI, configure:
set -a && . "$RATCHET_DIR/.env" 2>/dev/null && set +a
uv run --directory "$RATCHET_DIR" ratchet configure --llm-mode host-cli --host-agent codex
The --model flag is accepted for compatibility, but deterministic local mode
does not require a model alias.
Pipeline Outputs
Skills & Strategies — saved to pending dirs for review/install
Option 3 — Leave it running:
Return immediately. Do not print anything or ask further questions.
Handling Local Timeout (Exit Code 3)
If the pipeline command exits with code 3, the local poll timeout was hit
before the worker reached a terminal status. Parse the JSON output for
timeout.error details.
Use the AskUserQuestion tool to present these options:
Question: "The local pipeline timed out ({error message}). What would you like to do?"
Options:
"Run again — start a fresh pipeline run"
"Do nothing — exit without action"
Option 1 — Run again:
Re-execute the pipeline command from "Running the Pipeline" section.
Option 2 — Do nothing:
Return immediately. Do not print anything or ask further questions.
Post-Pipeline Workflow (MANDATORY)
The pipeline prints a JSON object with additionalContext on completion.
You MUST parse and follow the embedded workflow immediately — do NOT just report "pipeline complete".
Steps:
Parse run_id and project_id from the pipeline output JSON (additionalContext).
Run this command to get the detailed review workflow instructions:
