Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This skill guides technical founders through the lifecycle of building on Claude Managed Agents. It moves from initial scoping and v0 launch to iterative grading and scheduled deployment, whether the goal is an internal worker or a customer-facing product feature.
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Follow the documented setup, then try a first task.
Run the repository in Claude Code
Installs: Complete repository skill bundle under `.claude/skills/`, including `launch-your-agent` and its companion `wrap-up` skill. · Claude Code
Before you start
Claude Code installed and signed in.
An Anthropic API key for the user's own account is required from the launch step onward; the documented flow has the user create it at platform.claude.com → API keys and store it in a local `.env` file, never in chat.
Run from: launch-your-agent/
In a terminal, clone the repository, enter its directory, and start Claude Code.
git clone https://github.com/anthropics/launch-your-agent.git
cd launch-your-agent
claude
Inside the Claude Code session, enter `/launch-your-agent`. The skills under `.claude/skills/` are detected automatically while Claude Code runs in this repository folder.
A description of the job the managed agent should perform.
What a successful first version should accomplish.
Any essential inputs, outputs, or schedule requirements.
Use the launch-your-agent skill to interview me about `[the job I want an agent to perform]` and scope the smallest useful v0. For this first pass, produce only the proposed scope and definition-of-done rubric; do not launch anything.
An existing `./my-agent/` folder produced by launch-your-agent.
Optionally, `ANTHROPIC_API_KEY` stored in `my-agent/.env` for live status; without it, the skill continues in describe-only mode.
Use the wrap-up skill with the existing `./my-agent/` folder to refresh the overview page, recap the managed-agent primitives and run status, suggest the 1–2 most useful next upgrades, and perform the documented hygiene sweep. If `my-agent/.env` does not contain an API key, continue without live API status.
Claude
Copilot