Agent builds
Agent setups that ran. Fork one.
Short answer: download or fork real AI agent setups (CLAUDE.md files, Claude Code skills, prompts, MCP and swarm configs). Each build says what it does, which model it runs on, what a run costs and what it measured. Files are scanned so no keys get published.
Site-first article and explainer agents
Two Claude Code agents that publish to your own site first: one writes a data-backed article per run, the other turns articles and projects into pages that each answer one searched question.
#674This Week in Rust issue that featured the article agent's crates.io data piecethis-week-in-rust.orgon sourceAI outreach agent team: short-lived role agents on a schedule (free sample)
The layout of a Claude Code agent team that runs replies, inbox and numbers on a schedule. Each role wakes up, does one run, writes down what it learned, and exits. Sample of the paid kit.
962,969X views on the team's posts and replies, as counted on the team's public receipts pageagent-receipts-five.vercel.appon sourceWeekly report skill (free sample from the AI Builder Kit)
A ready-to-install Claude skill that turns rough notes into a short weekly status report: shipped, in progress, blocked, next week. One of the starter skills from Skillsmith in the AI Builder Kit.
sqlswarm: a cheat-proof head-to-head for coding agents
A setup for comparing coding agents fairly: same goal, same model, same clock, and a read-only grader (SQLite's sqllogictest) scored from outside so no agent can raise its score by editing tests.
throng: a swarm of coding agents verified by real builds and tests
Run many coding agents on one repo in parallel. Every build and test goes through one shared, cached engine, and you only get back a change the real test suite passed. Works from Claude Code over MCP.
What makes a good build
- It ran on a real job, and the card says what it did with a number and a source (a receipt).
- Someone else can run it: the README says what to install, what to change and what a run costs.
- No secrets. Use placeholders like <YOUR_API_KEY>. Publishing is blocked if the scanner finds a key.