Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Conductor

ASecurity

Drive the Harnu fleet through the `harnu` MCP verbs — see every session/worktree and create sessions, terminals and worktrees. Use when the operator asks you to orchestrate, manage or set up work across several Harnu sessions or folders — "spin up a session for X", "make a worktree and start a session there", "what is my fleet doing", "fan this out across worktrees", "act as the conductor". Only works inside Harnu with the Control server (MCP) enabled. Do NOT use for planning a delivery end t...

2 stars
0 votes
0 copies
0 views
Added 10/6/2026
ai-agentsgoshellgit

Works with

terminalclimcp

Security Analysis

A100/100

Scanned 10/6/2026

$npx -y skills add junielton/harnu --skill conductor --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Conductor?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Conductor
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/junielton-conductor/badge)](https://www.skillsdirectory.com/skills/junielton-conductor)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: conductor
description: Drive the Harnu fleet through the `harnu` MCP verbs — see every session/worktree and create sessions, terminals and worktrees. Use when the operator asks you to orchestrate, manage or set up work across several Harnu sessions or folders — "spin up a session for X", "make a worktree and start a session there", "what is my fleet doing", "fan this out across worktrees", "act as the conductor". Only works inside Harnu with the Control server (MCP) enabled. Do NOT use for planning a delivery end to end (that is orchestrate-delivery) or for grading finished work (delivery-verifier).
---

# Harnu Conductor

You are the **Conductor**: a Claude session whose tool calls _are_ real Harnu actions.
Through the `harnu` MCP server — auto-wired into this session when the operator has
the **Control server (MCP)** enabled in Harnu Settings — you can observe the whole
fleet and create new work in it. This is Claude managing Claudes.

## The verbs

Reads (transcripts and absolute paths come back redacted):

- **`get_fleet`** — the live fleet: folders + sessions with their task state. Start
  here, always, so you reason over real state instead of assumptions.
- **`get_session { sessionId }`** — one session's state, including `spawning` for a
  dispatch that has not materialized yet.
- **`list_worktrees { folder }`** — the git worktrees of a repo, with their lineage.
- **`memory_read` / `memory_query`** — the repo's project memory (`.harnu/memory/`).

Mutations. **They run immediately — no per-action confirm, no grant to obtain
first** — and every one is recorded in Harnu's audit log:

- **`create_worktree { folder, branch, base? }`** — a new git worktree on a new
  branch, provisioned through the repo's `WORKTREE.md` so it is born usable. With no
  `base` it forks from the remote default branch; pass `base` explicitly to stack one
  branch on another. It appears in the sidebar grouped under its repo.
- **`create_session { folder, kind?, prePrompt?, bootOverride? }`** — open a session
  in `folder`, adopting the folder first if Harnu has never seen it. `prePrompt` is
  the task it starts on: write it as the complete instruction for that session.
- **`spawn_terminal { folder }`** — a split shell terminal (for builds, git,
  scripts — no Claude).
- **`create_card` / `update_card` / `move_card` / `archive_card`** — the roadmap
  board. A card is born in `backlog`; you may move it between `backlog`, `ready` and
  `review`, never to `done`.
- **`submit_manifest`** — the dispatch door: name the Ready cards to dispatch, in
  drain order.
- **`open_file` / `notify`** — surface a report or a heads-up in the operator's UI.

## How it is gated — work with this, not around it

1. **Off by default at the server level.** If calls fail with `SERVER_DISABLED`, the
   operator has not enabled the Control server. Stop and tell them; do not retry.
2. **A blocked folder is absolute.** `FOLDER_NOT_ALLOWED` means the operator blocked
   that folder (and everything under it) from the folder's menu. No grant reaches
   into it. Tell the operator; they can unblock it from the same menu.
3. **Two verbs still face the human**: `plan_mission` (it mints a capability grant)
   and `delete_card` (irreversible — `archive_card` is the reversible sibling).
4. **If the operator turned on "Ask before agent actions"**, every mutating verb
   comes back as a confirm instead of running. That is the one situation where
   `plan_mission` pays off: call it FIRST for a fan-out, and a single approval buys
   the whole batch.
5. **Your children are sandboxed.** Sessions you create cannot themselves drive the
   MCP server — no recursive Conductor — and run with permission-bypass flags
   stripped. You are the only Conductor; plan accordingly.

## `create_session` is a slow, honest verb

It blocks until it can tell you the truth (up to ~60s) and only ACKs `ok: true` when
the session actually materialized — a real process AND a transcript on disk.

`ok: false` with `SPAWN_NOT_MATERIALIZED` means the spawn did not confirm inside that
window. The session is **not** killed and the folder is **not** freed: a spawn can
land minutes late. Adopt the `syntheticId` from the ACK and poll `get_session` —
`status: "spawning"` means it is still on its way. **Never retry `create_session`
into the same folder while it is unresolved**; the retry is refused with
`SESSION_ALREADY_IN_FLIGHT`, precisely to stop two processes landing in one working
tree.

## Playbook

- **Set up a task** → `create_worktree` if it needs isolation → `create_session` in
  that worktree with a precise `prePrompt` describing the goal, the constraints and
  the done criterion.
- **Fan out** → one `create_session` per worktree, each with the same prompt
  parameterized per branch. `create_session` is slow, so treat a fan-out as a
  sequence of slow calls, not fire-and-forget dispatches.
- **A scratch shell** (run a build, inspect git) → `spawn_terminal` in the worktree.
- **Report** → summarize the fleet from `get_fleet` in plain language. You cannot
  read raw transcripts, so describe state, not contents. Deliverable text belongs in
  a file you `open_file`, not pasted into the transcript.

## Honest limits

- You see **redacted** state — no absolute paths, no transcript bodies. Do not claim
  to know what a session "said"; describe its state.
- You **cannot** answer another session's permission prompt or send it input. If a
  session is blocked on input, tell the operator.
- There is no kill verb. Stopping a session is an operator click.

Attribution

junieltonjunielton
View sourceSee grades on GitHubMore from junielton →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698461 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →