Agent skill

Worktree Isolation

by autopus-ai in autopus-ai/autopus-adk

Worktree isolation for parallel agent execution in the pipeline

MITAuto-check passedAgent Workflows

Install Worktree Isolation

skills CLI
$ npx skills add autopus-ai/autopus-adk --skill worktree-isolation -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install autopus-ai/autopus-adk worktree-isolation --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/autopus-ai/autopus-adk.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.omp/skills/worktree-isolation .claude/skills/worktree-isolation && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
worktree-isolation
GitHub stars
110
Token cost
~1.1k tokens
SKILL.md length
616 words
Files
1
Skills in repo
11
Repo updated
First seen
Licence
MIT

At a glance

Worktree isolation for parallel agent execution in the pipeline

  • Works in 5 steps: normalize every owned and forbidden path… → reject absolute paths, traversal,… → detect exact overlap and… → …
  • Tasks that involve Git worktrees
  • SKILL.md covers Activation, Native Dispatch Contract, Ownership Preflight and Tool-Owned Lifecycle, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Worktree Isolation is an agent skill from autopus-ai/autopus-adk. Worktree isolation for parallel agent execution in the pipeline

Its SKILL.md is about 1.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts. Compatibility notes: omp

It sits in Agent Workflows, covering Git worktrees and Subagents. The repository describes itself as: Autopus-ADK is of the agents, by the agents. for the agents. Multi-model orchestration (consensus/pipeline/debate/fastest). Architecture-as-Code, Lore decision tracking… The licence is MIT.

When your agent uses it

  • Tasks that involve Git worktrees
  • Tasks that involve Subagents

Example prompts

  • “/worktree-isolation”

Requirements

  • Compatibility (from SKILL.md): omp

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. normalize every owned and forbidden path to a project-relative path;
  2. reject absolute paths, traversal, symlinks, nested ownership, and ambiguous globs;
  3. detect exact overlap and directory-prefix containment;
  4. serialize tasks that share a generated artifact, migration directory, package manifest, lockfile,
  5. put cross-task interfaces in top-level context before dispatch.

What it can do on your machine

Read from SKILL.md and the folder at commit fff509f. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

  • Compatibility

    omp

    From compatibility in the SKILL.md frontmatter.

Context cost

Worktree Isolation loads about 1.1k tokens when it runs. Until then it costs about 21 tokens; SKILL.md has 616 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~21
When it runs · the whole SKILL.md, loaded when a task matches
~1.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from autopus-ai/autopus-adk at commit fff509f, republished under its MIT licence (© autopus-ai). 616 words, ~1,127 tokens.

Download SKILL.mdSave it as .claude/skills/worktree-isolation/SKILL.md (or your agent's skills folder).
name
worktree-isolation
description
Worktree isolation for parallel agent execution in the pipeline
compatibility
omp

OMP Native Task Isolation

This skill defines isolation policy for implementation fan-out on OMP. The native task tool owns workspace materialization, patch/branch integration, and cleanup.

Activation

Use per-item isolation only when all conditions hold:

  • the current dynamic task schema exposes isolated;
  • OMP settings enable a non-none isolation mode;
  • the project is a git repository;
  • the item writes files;
  • ownership is disjoint from every concurrent item.

Do not request isolation for --solo, read-only work, overlapping ownership, shared migration numbering, or an environment where the field is absent. Missing required isolation is a blocker, not permission to imitate it with shell commands.

Native Dispatch Contract

Inspect the current dynamic schema before each wave. When it exposes batch mode, dispatch independent isolated work in one batch with top-level i and shared context. Each item uses a unique stable name, a complete assignment, the discovered isolation field enabled, the strict five-field receipt schema, and schemaMode: strict. Writing items omit the agent field so they run on the bundled task agent; select scout, reviewer, security-reviewer, or sonic only when that agent's built-in boundary is what the item needs. When batch mode is absent, dispatch the corresponding flat calls and reference one shared local:// context.

The parent must check dynamic availability before adding isolated or effort. isolated does not exist when task.isolation.mode = none, and effort does not exist when its setting is off. Omit either unsupported field rather than sending it from a static example.

Ownership Preflight

Before fan-out:

  1. normalize every owned and forbidden path to a project-relative path;
  2. reject absolute paths, traversal, symlinks, nested ownership, and ambiguous globs;
  3. detect exact overlap and directory-prefix containment;
  4. serialize tasks that share a generated artifact, migration directory, package manifest, lockfile, schema registry, or other mutable authority;
  5. put cross-task interfaces in top-level context before dispatch.

A maximum of five isolated implementation items may be active in one wave. Queue overflow by task id. This policy is stricter than OMP's session-wide semaphore and prevents excessive integration churn.

Show full SKILL.md (285 more words)Show less

Tool-Owned Lifecycle

For an isolated item, OMP:

  1. captures the parent baseline;
  2. creates the configured isolated workspace;
  3. runs the child in that workspace;
  4. captures a patch or commits a temporary task branch according to OMP settings;
  5. applies or merges the result into the parent through the task lifecycle;
  6. cleans the isolated workspace;
  7. preserves output, transcript, and patch metadata through agent:// and history://.

The Autopus parent must not run manual worktree creation, branch merge, cherry-pick, stash, reset, or removal commands. It verifies the returned changed_files, ownership boundary, blockers, and parent-tree result after OMP completes integration. Nested repositories are handled by OMP's nested patch lifecycle and must not be merged manually.

Sequential Work

A task is sequential when it depends on an earlier result, overlaps ownership, or shares a migration numbering lane. Dispatch it only after the prerequisite result is visible in the parent workspace. Do not use an isolated batch to hide a dependency edge.

Failure and Conflict Handling

  • A failed isolated run is terminal and not revivable because its workspace has been cleaned.
  • Keep the transcript and patch metadata as evidence.
  • If integration fails, stop the next wave and report the exact OMP lifecycle error.
  • Never bypass a failed patch/branch integration with destructive git commands.
  • A correction is a new explicitly named task with freshly declared ownership and context.
  • Cancel still-running jobs through hub with top-level i; preserve user-owned changes and unrelated jobs.

Receipt Verification

Every isolated worker returns exactly:

  • owned_paths
  • changed_files
  • verification
  • blockers
  • next_required_step

The main session rejects missing fields, out-of-scope changes, body dumps, secret material, or claims without observable verification. Integration is complete only after the parent sees the intended changes and the next deterministic gate passes.

© autopus-ai, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .omp/skills/worktree-isolation of autopus-ai/autopus-adk.

Open the folder on GitHubat commit fff509f

Compare with similar skills

Worktree Isolation next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Worktree Isolation compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Worktree Isolation this skillautopus-ai/autopus-adk110—~1.1kAutomated safety check: PassMIT
Vspawnvlinx-io/VelaTerm275—~2.5kAutomated safety check: PassMIT
Spec-Driven Development v2LichAmnesia/lich-skills234—~3.1kAutomated safety check: PassMIT
Subagent Coordinatorflyxl/datazen114—~936Automated safety check: PassGPL-3.0
Batch Orchestrationrohitg00/pro-workflow2.9k—~1.2kAutomated safety check: PassNone
PmApra-Labs/apra-fleet101—~5.4kAutomated safety check: PassCustom licence

Similar skills

  • Vspawn

    vlinx-io/VelaTerm

    Explicitly spawn a standalone child session under the current vlx-term session, passing the task in as its first message (mirrors spawntask).

    275 GitHub stars~2.5k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed
  • Spec-Driven Development v2

    LichAmnesia/lich-skills

    Organizes long-running agent work into a Project, Sprint and Task hierarchy with per-task state files, isolated worktrees, review loops and script-checked rules.

    234 GitHub stars~3.1k tokensUpdated 4 mo ago
    Agent WorkflowsAuto-check passed
  • Orchestrate multi-track parallel feature development with subagents and git worktrees.

    114 GitHub stars~936 tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Batch Orchestration

    rohitg00/pro-workflow

    Decompose large-scale changes into independent units and spawn parallel agents in isolated worktrees.

    2.9k GitHub stars~1.2k tokensUpdated 10 days ago
    Agent WorkflowsAuto-check passed
  • Pm

    Apra-Labs/apra-fleet

    Project Manager skill. An agent skill from Apra-Labs/apra-fleet.

    101 GitHub stars~5.4k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Codex CLI

    kortix-ai/suna

    Drive OpenAI's Codex CLI (codex exec) as a non-interactive coding sub-agent from inside Claude Code.

    20k GitHub stars~1.6k tokensUpdated today
    Agent WorkflowsAuto-check passed

More from autopus-ai/autopus-adk

All 11 skills in this repo
  • Auto Idea

    autopus-ai/autopus-adk

    아이디어 브레인스토밍 — 멀티 프로바이더 토론과 ICE 평가로 아이디어를 정리합니다. An agent skill from autopus-ai/autopus-adk.

    110 GitHub stars~2.4k tokensUpdated today
    Auto-check passed
  • Auto QA

    autopus-ai/autopus-adk

    QAMESH project QA mesh — plan, run, report, and publish deterministic QA evidence

    110 GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Auto Setup

    autopus-ai/autopus-adk

    프로젝트 컨텍스트 생성 — 코드베이스를 분석하고 ARCHITECTURE.md 및 .autopus/project 문서를 생성합니다

    110 GitHub stars~1.5k tokensUpdated today
    Auto-check passed
  • Auto Sync

    autopus-ai/autopus-adk

    문서 동기화 — 구현 이후 SPEC, CHANGELOG, 문서를 반영합니다. An agent skill from autopus-ai/autopus-adk.

    110 GitHub stars~1.1k tokensUpdated today
    Auto-check passed
  • Ax Annotation

    autopus-ai/autopus-adk

    @AX code annotation workflow skill for agent-driven tag application

    110 GitHub stars~1k tokensUpdated today
    Auto-check passed
  • Browser Automation

    autopus-ai/autopus-adk

    터미널 환경 자동 감지 브라우저 자동화 스킬 — AI 에이전트가 직접 웹 페이지를 조작하고 검증. An agent skill from autopus-ai/autopus-adk.

    110 GitHub stars~2.1k tokensUpdated today
    Auto-check passed

Questions about Worktree Isolation

What does Worktree Isolation do?

Worktree isolation for parallel agent execution in the pipeline. Worktree Isolation is an agent skill from autopus-ai/autopus-adk.

When should I use Worktree Isolation?

Worktree Isolation fits situations like: tasks that involve Git worktrees; tasks that involve Subagents.

How do I install Worktree Isolation in Claude Code?

Run `npx skills add autopus-ai/autopus-adk --skill worktree-isolation -a claude-code`. Or copy the skill folder (.omp/skills/worktree-isolation in autopus-ai/autopus-adk) into .claude/skills/worktree-isolation in your project. Claude Code loads it when a task matches its description.

How do I install Worktree Isolation in Codex?

Run `npx skills add autopus-ai/autopus-adk --skill worktree-isolation -a codex`. Or copy the skill folder (.omp/skills/worktree-isolation in autopus-ai/autopus-adk) into .agents/skills/worktree-isolation in your project. Codex loads it when a task matches its description.

Can I use Worktree Isolation in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add autopus-ai/autopus-adk --skill worktree-isolation -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/worktree-isolation, .gemini/skills/worktree-isolation, .github/skills/worktree-isolation and .opencode/skills/worktree-isolation in your project.

What does Worktree Isolation need to run?

SKILL.md names no scripts, command-line tools or credentials: Worktree Isolation is instructions for the agent only. Compatibility (from SKILL.md): omp.

Does Worktree Isolation access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Worktree Isolation safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Worktree Isolation use?

Worktree Isolation is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Worktree Isolation use?

About 1.1k tokens (SKILL.md is roughly 4.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Worktree Isolation?

Skills that share tags, products or a category with Worktree Isolation: Vspawn (vlinx-io/VelaTerm, 275 stars), Spec-Driven Development v2 (LichAmnesia/lich-skills, 234 stars), Subagent Coordinator (flyxl/datazen, 114 stars) and Batch Orchestration (rohitg00/pro-workflow, 2.9k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Worktree Isolation?

autopus-ai (a GitHub organization) maintains it in autopus-ai/autopus-adk, which has 110 GitHub stars. The repository holds 11 skills in this directory. The repository was last updated on October 9, 2026.

Source: autopus-ai/autopus-adk on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.