agentleFS
Sign inStart free
tmdytmdy/CodeDeck/skills/delegate-with-codedeck/SKILL.md

delegate-with-codedeck

Skill2 starsChanged 26 days agoMITView on GitHub

Delegate a coding or research task to a Codex or Claude Code CLI managed by a local CodeDeck Profile. Use when the user asks to consult, invoke, compare, or hand work to a third-party Codex/Claude profile through CodeDeck, or asks to continue or inspect an existing CodeDeck delegated session.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 1 is distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use scripts/invoke-codedeck.ps1 for every operation.

    line 8
  2. Very specific:

    If CODEDECK_DELEGATED=1, do not start or control unrelated tasks.

    line 13
  3. Very specific:

    Run profiles when the user did not provide an exact Profile key.

    line 17
  4. Very specific:

    Run the task with the selected Profile.

    line 20
  5. Very specific:

    Use -Orchestrator only when the root caller needs a tracked parent session that may launch one bound delegated child.

    line 21
  6. Very specific:

    Use -Model and -ReasoningEffort for run-only overrides.

    line 24
  7. Very specific:

    Add -ShowTerminal to run only when the user requests visibility or manual takeover.

    line 54
  8. Distinctive:

    Read references/protocol.md only when handling errors or integrating a new host.

    line 61

≈ “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

Delegate With CodeDeck

Use scripts/invoke-codedeck.ps1 for every operation. Do not reconstruct the pipe protocol or read CodeDeck configuration files directly.

Workflow

  1. If CODEDECK_DELEGATED=1, do not start or control unrelated tasks. The only permitted commands are policy for this child binding, plus status-parent, continue-parent, and interrupt-parent for the server-bound parent. These commands never accept a target session ID.
  2. Run profiles when the user did not provide an exact Profile key. Present only the returned safe summaries and select a Profile based on the user's request.
  3. Run the task with the selected Profile. Pass the prompt through -Prompt. Use -Orchestrator only when the root caller needs a tracked parent session that may launch one bound delegated child; use -Cwd only when the task needs a specific existing project directory. Use -Model and -ReasoningEffort for run-only overrides. From a monitored parent, use -WaitMs 0 and select -ReturnPolicy once|continuous|manual when the parent should finish its turn instead of blocking for the child.
  4. Interpret the returned status:
    • completed: use assistant_text as the delegated answer.
    • starting or running: retain session_id and call wait.
    • awaiting_input: surface question; never guess or auto-approve a permission request. Use continue only with an explicit answer.
    • failed: report the structured error and relevant terminal_tail.
  5. Retain session_id for follow-up requests in the same conversation.

Commands

& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command profiles
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command run -Profile "codex::name" -Cwd "C:\repo" -Prompt $task
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command run -Profile "codex::name" -Model "gpt-5.6" -ReasoningEffort high -ReturnPolicy continuous -WaitMs 0 -Prompt $task
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command run -Profile "codex::name" -Orchestrator -ShowTerminal -WaitMs 0 -Prompt $task
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command wait -Session $sessionId
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command continue -Session $sessionId -Prompt $answer
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command status -Session $sessionId
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command interrupt -Session $sessionId
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command cancel -Session $sessionId
# Only inside the delegated child:
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command policy -ReturnPolicy manual
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command status-parent
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command continue-parent -Prompt $messageForParent
& "<skill-dir>/scripts/invoke-codedeck.ps1" -Command interrupt-parent

Add -ShowTerminal to run only when the user requests visibility or manual takeover. The default wait is ten minutes; use -WaitMs when the host tool has a shorter deadline. interrupt/interrupt-parent sends ESC to the peer's current turn; it does not close the terminal or destroy the binding. For an active A/B exchange, use manual: the child explicitly checks the parent and calls continue-parent instead of relying on automatic completion delivery.

Read references/protocol.md only when handling errors or integrating a new host.

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.

Check your setup, free