engine

Thorough analysis of code, architecture, or topics — produces a structured research report. Supports goal-based modes: Explore (general research), Audit (risk-focused critique), Improve (actionable suggestions), Custom (user-defined lens). Triggers: "analyze this code", "deep dive into", "research this topic", "investigate how X works", "audit this", "critique this", "suggest improvements", "find risks in".

Structured ideation and trade-off analysis for design and architecture decisions. Triggers: "brainstorm ideas", "explore this problem", "think through trade-offs", "challenge assumptions", "discuss architecture".

Executes routine maintenance and cleanup tasks from a structured task queue. Triggers: "do some chores", "housekeeping tasks", "small cleanup tasks", "work through a task queue".

Drives feature implementation following structured development protocols. Triggers: "implement this feature", "build this", "write the code", "TDD implementation", "execute the plan".

Iterative prompt and schema refinement using TDD methodology for LLM workloads. Triggers: "refine the prompt", "improve extraction", "iterate on schema", "prompt TDD", "tune the model".

Reviews and validates work across sessions for consistency and correctness. Triggers: "review session work", "validate debriefs", "approve session reports", "end-of-day review".

Designs and writes test cases for code correctness and regression prevention. Supports goal-based modes: Coverage (gap-filling), Hardening (edge cases & stress), Integration (boundaries & contracts), Custom (user-defined lens). Triggers: "write tests", "design test cases", "verify edge cases", "catch regressions", "test this feature", "harden this", "integration tests".

Keeps documentation in sync with code changes and project state. Triggers: "update documentation", "patch the docs", "sync docs with code changes", "update architecture docs".

Creates, edits, or promotes skills — scaffolds SKILL.md and assets to project-local or shared engine. Triggers: "create a new skill", "edit a skill", "scaffold a skill", "promote a skill", "add a project skill", "create a new command".

Smart assistant for the workflow engine — answers questions, invokes scripts, and navigates docs. Triggers: "engine help", "how does the engine work", "run an engine command", "what scripts are available".

Interactive agent fleet designer - configure your multi-agent workspace. Triggers: "configure fleet", "setup agents", "multi-agent workspace", "fleet layout".

Checks and retrieves completed research results from Gemini Deep Research. Triggers: "fulfill a research request", "check research results", "poll Gemini for research results".

Full research cycle — refines query, calls Gemini Deep Research, polls, delivers report. Combines /research-request + /research-respond. Triggers: "research this topic", "run a research cycle", "call Gemini Deep Research".

Captures and persists current session context for later restoration. Triggers: "dehydrate this session", "serialize session state", "save session snapshot", "pause and resume later".

Re-initializes session context after context overflow restart. Internal skill - invoked by session.sh restart.

Refines existing documentation for clarity, accuracy, and structure. Triggers: "restructure the docs", "consolidate documentation", "improve doc readability", "fix stale documentation", "clean up docs".

Posts a research request for async fulfillment by Gemini Deep Research. Triggers: "post a research request", "create a research question", "queue a research task for Gemini".

Generates a progress report summarizing work across sessions. Triggers: "summarize progress", "generate progress report", "what was done today", "status update across sessions".

Creates semi-permanent situational documents in project docs folders. Lightweight skill — no session, no logs, just focused writing. Triggers: "write up", "writeup".