@developerz.ai/aitm

🤖 @developerz.ai/aitm

Autonomous task orchestrator. Goal in, merged PRs out.

Give aitm a sentence, walk away, come back to a stack of merged pull requests. It runs a Planner → Worker → Reviewer loop against a real repo using the Vercel AI SDK and OpenRouter, and ships the work as PR-sized commits with CI gating and review-comment handling baked in.

💡 Spiritual successor to the (deprecated) developerz-ai/claude-task-master — same idea, rebuilt on the AI SDK + OpenRouter: provider-agnostic, concurrent PR groups, MCP client support, smaller surface area.

⚡ Install

npm  install -g @developerz.ai/aitm
bun  install -g @developerz.ai/aitm
deno install -A npm:@developerz.ai/aitm

The package is scoped; the installed command is just aitm.

🚀 Quickstart

export OPENROUTER_API_KEY=sk-or-...
cd path/to/your/repo
aitm start "add JWT auth to /login" --max-prs 3

aitm plans the goal into up to 3 PR-sized groups, opens a branch per group, works through them in parallel, opens each PR, watches CI, addresses review comments, and auto-merges. Want a human gate?

aitm start "migrate Mongo → Postgres" --no-automerge
# ... review the PR in your browser ...
aitm merge-pr

🧠 How it works

Role	Responsibility
Orchestrator	Top-level agent; drives the run group-by-group
Planner	Goal in → ordered PR groups (each a list of tasks) out
Worker	One PR group in → commits + an opened PR out
Reviewer	Review comments in → follow-up commits out

Subagents are wired with the subagents-as-tools pattern (isolated context windows, focused prompts, natural parallelism), built on @developerz.ai/ai-claude-compat.

🎯 Use cases

What you type	What aitm does
aitm start "add password reset flow"	Splits into schema + endpoint + email + tests, one PR per slice
aitm start "rename Logger to Tracer everywhere" --max-prs 1	Single sweeping PR, full test pass before merge
aitm start "add tests for src/billing/* until 90% coverage"	Iterates until the coverage target hits, or the session cap
aitm start "bump zod to v4 and fix all type errors"	Bumps, fixes, runs tests, opens PR; conflicts surface as blocked

⚙️ Configuration

User config lives at ~/.aitm.json; per-project overrides at .ai-task-master/config.json:

aitm config set models.smart  anthropic/claude-opus-4.7
aitm config set models.coding anthropic/claude-sonnet-4.6
aitm config set models.fast   openai/gpt-5-mini
aitm config set autoMerge true --project
aitm config list

• Provider: any OpenAI-compatible endpoint via one credential — OpenRouter by default, or set baseURL to run on z.ai GLM, a self-hosted gateway, etc. No Anthropic SDK. Profiles switch the whole provider in one command:

  aitm profile add z.ai --preset zai --api-key "<your z.ai key>"
  aitm profile use z.ai     # ✅ verified end-to-end on z.ai GLM (glm-5.2 / glm-5-turbo)
  aitm profile use openrouter
  

  See providers and aitm profile.

• Coding style: aitm reads your repo's CLAUDE.md / AGENTS.md and feeds it to subagents as a style signal (the provider stays OpenRouter).

• MCP: aitm is an MCP client — declare mcpServers in config and their tools mount into the subagent tool surfaces.

🛠 Requirements

• OPENROUTER_API_KEY in the environment.
• The gh CLI, authenticated (PRs, CI status, reviews).
• Node ≥ 20, Bun, or Deno ≥ 1.40 — ESM, no runtime lock-in.

License

MIT · source & full docs