ShipSafe · Bootcamp

Agent Bootcamp

Ship safely with coding agents (Cursor, Claude Code, Codex and others). Rules, skills, MCP permissions, review habits and deploys. 5 modules, 15 lessons, 15 labs (6 flagship).

Start the first lab Free, no signup. Progress stays in this browser.

1. 01
   Rules and context files

   AGENTS.md, CLAUDE.md, .cursor/rules: what the agent always sees. 3 lessons.

   • Start with AGENTS.md (or the equivalent) flagship lab
   • What belongs in rules vs what belongs in chat flagship lab
   • When the context window compacts lab
2. 02
   Skills

   SKILL.md folders, when to install them, how to review them. 3 lessons.

   • What a skill is lab
   • Review a skill before you trust it flagship lab
   • Use ShipSafe's reviewed list lab
3. 03
   MCP servers and tool permissions

   Least privilege, secrets in tool context, prompt injection via tools. 3 lessons.

   • Least privilege for MCP tools flagship lab
   • Secrets in tool output become context lab
   • Prompt injection through tools and pages lab
4. 04
   Reviewing and testing agent output

   Diff review, tests, evals before you merge. 3 lessons.

   • Read the diff like a hostile patch flagship lab
   • Make the agent write or run tests lab
   • Check every package the agent added lab
5. 05
   Agent-safe deploys and AI bots

   Preview isolation, env hygiene, and bots that read your docs. 3 lessons.

   • Agents and env vars lab
   • Your docs are training and search fodder lab
   • Run the checklist before you call it done flagship lab