Agents

Watches quest progress via task metadata during fellowship execution. Detects stuck quests (no phase change in extended periods), scope drift (implementation diverging from plan), and file conflicts between parallel quests. Reports issues to Gandalf for intervention.

The workhorse agent that executes the quest lifecycle. Uses the fellowship CLI binary for gate management, status checks, and phase transitions. Each quest runner operates in its own isolated git worktree, ensuring parallel quests never interfere with each other.

Attacks the implementation before it reaches review. Analyzes the quest diff for failure modes, writes targeted test cases using the project's existing test framework, runs them, and delivers a severity-ranked findings report. Critical and High findings block the Review gate until addressed.

Autonomous research agent for questions that need investigation but not code changes. Explores the codebase, reads documentation, and optionally spawns a fresh adversarial validator subagent to verify findings. Produces structured reports with confidence levels.