ai-plugins

Make a repository worktree-ready: goal-driven per-worktree isolation plus an enforcement guard that protects the main checkout.

John's standard engineering regime as a reusable, stack-agnostic starting point for any serious project.

Portable guardrails for building, evaluating, and delivering LLM and agentic systems.

Capture sanitized eval cases from bad or borderline AI-assistant behavior and submit them to this marketplace.

Babysit a pull/merge request to completion — watch CI, respond to review feedback, and merge — across GitHub, Forgejo, and GitLab.

John's personal development workflow discipline: test-first implementation, scoped verification, debugging, review handling, and skill authoring.

Git-backed task boards for coding agents: branch-local .tasks storage, CLI, stdio MCP, dry-run-first scaffolding with explicit approval, and read-only dashboard workflows.