praxis

Write minimal implementation to make failing tests pass. Use this agent when the implement skill dispatches an implementation task from a yak with agent-type 'agent-impl'. The agent writes implementation ONLY — never modifies test files. Applies a YAGNI simplification pass after tests pass. Reports GREEN gate status.

Execute non-TDD phase tasks like schema migrations, infrastructure setup, or configuration. Use this agent when the implement skill dispatches a task from a yak with agent-type 'no-test'. The agent executes the task as described and verifies the acceptance gate.

Fix implementation after validation finds failures. Use this agent when the implement skill creates a remediation yak with agent-type 'agent-remediate'. The agent fixes implementation code ONLY — never modifies test files. Receives failure output from the validation agent. Escalates to user after 2 failed attempts.

Write failing tests for a TDD phase. Use this agent when the implement skill dispatches a test-writing task from a yak with agent-type 'agent-test'. The agent writes tests ONLY — never implementation code. It receives a self-contained task context with file paths, test spec, and test command. Reports RED gate FAIL if tests pass immediately.

Run the full test suite and report results without modifying any files. Use this agent when the implement skill dispatches a validation task from a yak with agent-type 'agent-validate'. The agent runs tests, type-checks, and linting — all must exit 0 for the VALIDATE gate to pass. Reports only, never fixes.

Implement phase of RPI methodology. Executes a task-graph-driven implementation using isolated agents with strict RED/GREEN/VALIDATE gate enforcement. Use when executing an implementation plan from the plan-tasks skill.

Plan phase of RPI methodology. Consumes research artifact or topic and produces compact implementation plan with test specs and Agent Context blocks. Activates naturally inside plan mode for non-trivial work, or invoke explicitly with /plan-tasks. Use after research or to start planning a non-trivial feature.

Refactoring process with test safety. Invoke immediately when user or document mentions refactoring, or proactively when code gets too complex or messy.

Post-session skill to reflect on what was built and produce improvement proposals. Reads recent git history, artifacts, and context files to extract learnings for skills, CLAUDE.md, hooks, and plan templates. Use after any substantive session as a post-mortem or retrospective.

Research phase of RPI methodology. Spawns parallel subagents for codebase exploration AND web/pattern research, then synthesizes findings for user review. Produces a temporary research artifact at .light/sessions/ and transitions into plan mode. Use before implementing non-trivial features to understand what already exists.