Batch Migration Orchestrator

Use when the user wants a large, mechanical change split into many independent units and executed in parallel with isolated workers and PRs.

Security Scan

The skill is internally consistent with its stated purpose (planning and running large parallel codebase migrations) but is fairly high-level and leaves important authentication and safety details to the agent/runtime — review how it will get repo access and how PRs will be created before running it.

Purpose & Capability

Name, description, and runtime instructions align: the skill describes splitting a large change into isolated units, validating each, and producing PRs/patches. It does not ask for unrelated resources or permissions.

Instruction Scope

The SKILL.md instructs the agent to inspect the repository, split work, launch parallel workers, and collect PRs/patches. This is expected for a migration orchestrator, but the instructions are high-level and open-ended: they do not specify how to authenticate to remotes, how to obtain or limit write access, nor do they include explicit safeguards about handling secrets that may appear in repo content. The agent running this will need to read repository files and may push branches/PRs using whatever VCS credentials are available.

Install Mechanism

Instruction-only skill with no install steps and no code files, so nothing will be written to disk by the skill itself during installation.

Credentials

The skill declares no required environment variables or credentials, but its workflow implies needing VCS credentials (e.g., to push branches or create PRs) and access to the repository on disk. Because those credentials are not declared, you should verify which runtime credentials the agent will use and that they have appropriately limited scope (repo-level or machine-level Git credentials, tokens for GitHub/GitLab with minimal scopes).

Persistence & Privilege

always is false and there is no indication the skill requests permanent or elevated platform-level privileges. It does not modify other skills or system-wide settings in the provided instructions.

Assessment

This skill appears coherent for orchestrating large code migrations, but it is high-level and relies on the agent/runtime to perform repository reads and to push branches/PRs. Before installing or invoking it: 1) Confirm which credentials the agent will use to access remotes (GitHub/GitLab tokens, SSH keys) and restrict their scope to only the needed repositories and actions. 2) Prefer a dry-run mode so you can review planned changes and generated PRs before merging. 3) Ensure CI/tests run on worker branches and that workers cannot write overlapping files. 4) Scan the repository for accidental secrets (API keys, passwords) to avoid exfiltration as part of broad inspections. 5) Limit concurrency or require human approval for high-impact units. If you need the skill to create PRs on your behalf, explicitly provision a dedicated machine/token with least privilege rather than reusing broad credentials.

Use this skill for sweeping refactors, migrations, bulk renames, or repetitive codebase-wide edits.

Goal

Turn one large request into a safe parallel rollout with independent units, validation, and mergeable outputs.

Workflow

1. Inspect the repo to find every affected area, convention, and dependency.
2. Split the change into independent units by module, directory, or subsystem.
3. Define an end-to-end verification recipe for each unit.
4. Launch workers in parallel only when their write scopes do not overlap.
5. Track completion, collect PRs or patches, and summarize failures.

Guardrails

• Do not parallelize tightly coupled work.
• Do not let workers edit the same files.
• Require each worker to validate its own unit.
• Prefer worktree or isolated branch execution when git is available.

Inputs

• Migration target
• Scope boundaries
• Verification method
• Branch/PR expectations

Source Provenance

Derived from Claude Code bundled skill logic in src/skills/bundled/batch.ts.