A structured approach to feature planning that auto-sizes itself based on complexity. Small changes skip straight to implementation, while larger features get full specs with requirement IDs, architecture decisions, and atomic task breakdowns. The workflow is always Specify then Execute, with optional Design and Tasks phases in between that only kick in when scope demands it. Keeps project state, decisions, and blockers in persistent memory across sessions. Works for both greenfield projects and mapping existing codebases. The context loading strategy is thoughtful, keeping base load around 15k tokens and using sub-agents for research and parallel task execution. Opinionated but flexible enough to get out of your way when you're just fixing a bug.
npx skills add https://github.com/tech-leads-club/agent-skills --skill tlc-spec-driven