agentleFS
qf-studioqf-studio/navigator/skills/nav-pilot/SKILL.md

nav-pilot

Skill355 starsChanged 13 months agoMITView on GitHub

Dispatch a Navigator task doc to Pilot as a labeled GitHub issue. Use when user says "dispatch TASK-XX to Pilot", "hand off to Pilot", "send to Pilot", or "queue for Pilot". One-way handoff (Navigator authors the spec, Pilot executes autonomously).

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 4 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    DO NOT invoke: when:

    line 25
  2. Very specific:

    No task doc exists yet (tell them to create one with nav-task first)

    line 28
  3. Distinctive:

    If the file does not exist, stop and tell the user to author it first.

    line 37
  4. Distinctive:

    Read the doc and extract:

    line 39
  5. Very specific:

    Read pilot from .agent/.nav-config.json:

    line 46
  6. Very specific:

    If pilot.enabled is false, stop and tell the user to enable it.

    line 60
  7. Very specific:

    --dry-run mode: if the user asked for a dry run (or says "dry-run" here), print the exact gh command and STOP.

    line 90
  8. Very specific:

    Run the gh issue create command.

    line 95
  9. Very specific:

    not authenticated → tell the user to run gh auth login

    line 99
  10. Very specific:

    no git remote / repo not found → ask for pilot.repo in config

    line 101
  11. Very specific:

    Add the issue URL under ## Refs: - Pilot issue: <url>

    line 106
  12. Very specific:

    Update the status line to Status: 🚀 Dispatched to Pilot.

    line 108
  13. Very specific:

    Update Last Updated: to today.

    line 109
  14. Distinctive:

    If the knowledge graph is enabled, the Edit will sync via the normal PostToolUse hook — no extra action needed.

    line 124

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Navigator → Pilot Dispatch Skill

Hand a finished Navigator task doc off to Pilot by creating the pilot-labeled GitHub issue Pilot polls, then record the issue back into the doc.

This skill does NOT validate the doc. Spec checking is Pilot's job (its spec_validator). nav-pilot only resolves the doc, creates the labeled issue, and records the result. It consumes what nav-task authored — it does not author or archive docs itself.

When to Invoke

Invoke when the user says:

  • "Dispatch TASK-XX to Pilot"
  • "Hand this off to Pilot" / "send to Pilot"
  • "Queue TASK-XX for Pilot" / "create the Pilot issue"

DO NOT invoke when:

  • User is authoring/archiving a task doc (that's nav-task)
  • User wants Pilot worker status (out of scope — this skill is one-way)
  • No task doc exists yet (tell them to create one with nav-task first)

Execution Steps

Step 1: Resolve the Target Task Doc

  • Explicit ID ("TASK-07"): use .agent/tasks/TASK-07-*.md.
  • No ID: pick the active/in-progress task. If ambiguous, list candidates (ls .agent/tasks/*.md) and ask which one.
  • If the file does not exist, stop and tell the user to author it first.

Read the doc and extract:

  • Title = the first H1 (# TASK-XX: <name>), stripped of the TASK-XX: prefix if present (or keep the full H1 — match what Pilot expects).
  • Body file path = the doc itself (passed via --body-file, never inlined).

Step 2: Load Pilot Config

Read pilot from .agent/.nav-config.json:

"pilot": {
  "enabled": true,
  "label": "pilot",
  "repo": null
}
  • label (default "pilot"): the label Pilot polls for.
  • repo (default null): target owner/name. When null, gh uses the current git repository's origin — do NOT pass --repo.

If pilot.enabled is false, stop and tell the user to enable it.

Step 3: Build the Command

gh issue create \
  --title "<H1 title>" \
  --label "<pilot.label>" \
  --body-file ".agent/tasks/TASK-XX-<slug>.md" \
  [--repo "<pilot.repo>"]   # only when pilot.repo is set

Step 4: Pre-flight Confirmation (ToM Checkpoint — high-stakes)

Creating a GitHub issue is an outward-facing action. ALWAYS confirm first:

About to dispatch to Pilot:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Repo:  <pilot.repo or "current origin (gh default)">
Title: <H1 title>
Label: <pilot.label>
Body:  .agent/tasks/TASK-XX-<slug>.md (<N> lines)

Command:
  gh issue create --title "..." --label "pilot" --body-file "..."
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Create this issue? [Enter to dispatch / "dry-run" / corrections]

--dry-run mode: if the user asked for a dry run (or says "dry-run" here), print the exact gh command and STOP. Do not execute.

Step 5: Dispatch

Run the gh issue create command. Capture the returned issue URL from stdout (gh issue create prints the URL on success).

If gh fails:

  • not authenticated → tell the user to run gh auth login
  • label missing on the repo → offer gh label create pilot
  • no git remote / repo not found → ask for pilot.repo in config

Step 6: Record Back Into the Doc

Edit the task doc:

  • Add the issue URL under ## Refs: - Pilot issue: <url>
  • Update the status line to **Status**: 🚀 Dispatched to Pilot.
  • Update **Last Updated**: to today.

Step 7: Confirm

✅ Dispatched to Pilot

Issue:  <url>
Label:  pilot
Task:   TASK-XX — <title>

Recorded back into .agent/tasks/TASK-XX-<slug>.md (status → 🚀 Dispatched).
Pilot will pick it up on its next poll of `pilot`-labeled issues.

If the knowledge graph is enabled, the Edit will sync via the normal PostToolUse hook — no extra action needed.

Error Handling

SituationResponse
No task doc foundAsk which task; suggest nav-task to create one
pilot.enabled: falseStop; tell user to enable in .nav-config.json
gh not authenticatedPrompt gh auth login
pilot label missingOffer gh label create pilot
No remote / repo unknownAsk user to set pilot.repo

Success Criteria

  • Correct task doc resolved and H1 used as the issue title
  • Pre-flight confirmation shown before any network call
  • --dry-run prints the command without executing
  • Issue created with the pilot label
  • Issue URL recorded back into the doc; status → 🚀 Dispatched

Notes

  • One-way by design: monitoring Pilot workers and pulling results back are out of scope (see .agent/tasks/TASK-54-nav-pilot-dispatch-skill.md).
  • Validation is intentionally absent — Pilot's spec_validator owns that.
  • Pairs with nav-task (authors the doc) → nav-pilot (dispatches it).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.