fleet

Autonomous TDD implementation agent. Picks up a story/spec, writes tests from acceptance criteria, implements until tests pass, self-checks for stubs, updates the spec, and commits. Spawned by fleet-run for parallel execution.

Post-build verification agent. Checks spec compliance, stub contamination, code quality, security, and regression. Spawned by fleet-run after each build wave.

Security audit agent. Checks for OWASP top 10 vulnerabilities, missing authentication, injection vectors, hardcoded secrets, and missing access control. Spawned on demand or as part of fleet-review.

Synchronization agent. Diffs repository spec state against Linear and Notion, reconciles differences. Repo always wins. Can be scheduled to run on a cron for continuous sync.

Honest brownfield codebase audit. Decomposes a repo into assessable modules, checks each for real implementation vs stubs/mocks/TODOs, grades test quality, scans infrastructure, and produces a truthful assessment report. Requires fleet-discover manifest.

Autonomous TDD build loop for a single story/spec. Reads the spec, writes tests from acceptance criteria, implements until tests pass, self-checks for stubs, updates the spec, and commits. Works with both BMAD stories and Fleet-generated specs. Called by fleet-run for parallel execution.

Maps an unknown brownfield repository. Detects languages, frameworks, databases, ORMs, test tools, CI, hosting, and monorepo structure. Produces a machine-readable manifest (_fleet/manifest.json) and a human-readable summary (_fleet/manifest.md). Scales detection strategy based on repo size. Use when onboarding a codebase that has no Fleet or BMAD artifacts.

Full project health check. Runs all audits, checks infrastructure, verifies spec completion against code, checks Linear/Notion sync drift, and produces a unified report. The one command to understand project state.

Process enforcement for Fleet. Installs Claude Code hooks that prevent process violations — no code without specs, no merge without tests, auto-formatting, session context loading. Run by fleet-infra during setup, or manually to install/update hooks.