Agent Claude Code
Focused implementation agent. Use when a clear specification exists and needs precise code manifestation. The Chariot drives implementation forward without deviation—precision is its virtue. Invoke when: plan approved, spec clear, need focused coding without feature creep.
Agent Claude Code
Use when a major project step completes and needs review against the original plan and coding standards. Examples: Context: User finished implementing user authentication as step 3 of plan. user: "I've finished implementing the user authentication system as outlined in step 3 of our plan" assistant: "Let me use the…
Agent Claude Code
Resource governance agent. Use to track scope creep, token usage, and project drift. The Emperor's gaze is fixed on the finite—budgets, time, focus. Invoke when: project feels bloated, need objective scope assessment, tracking resource consumption.
Agent Claude Code
Use for local git operations only — read, status, diff, log, add, commit, branch, fetch, and worktree. Does NOT push. A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Use for git push operations only. Operator confirmation is REQUIRED for every push. A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Wisdom extraction agent. Use after project completion to distill enduring lessons. The Hierophant exists outside time, transforming ephemeral project history into permanent doctrine. Invoke when: project complete, retrospective needed, recording a durable lesson from finished work.
Agent Claude Code
Use for worktree implementation work — editing source, running scoped Bash commands, and committing changes inside a parent-specified scope. A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Use for Atlassian/Jira write operations via scoped Atlassian MCP write tools — create issue, transition status, add comment, edit fields. Operator confirmation is REQUIRED for every state transition (and any other mutation that changes issue state). Jira access uses runtime-discovered Atlassian MCP write tools; the…
Agent Claude Code
Use for read-only Atlassian/Jira inspection — fetching issues, comments, sprints, and project metadata via Atlassian MCP read tools. Performs no mutations. Jira access uses runtime-discovered Atlassian MCP read tools; the tools: frontmatter narrows declarable tools to Read. Returns structured JSON.
Agent Claude Code
Conflict synthesis agent. Use when code and critique exist in tension—thesis and antithesis need synthesis. Justice weighs both sides with equal honor and produces refined solutions that respect both positions. Invoke when: code review identified issues, refactoring needed, disagreements require arbitration.
Agent Claude Code
Integration harmony agent. Use when modules exist but connections between them need review. Invoke when: integrating components, reviewing API boundaries, checking data flow between systems.
Agent Claude Code
Use for creating and editing pull requests via gh pr create, gh pr edit, gh pr view, gh pr diff, and gh pr list. Does NOT merge or mark ready (use pr-merger for that). A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Use for gh pr merge and gh pr ready only. Operator confirmation is REQUIRED for every merge or ready-mark. A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Emotional state monitor. Use to sense when a project is stuck, frustrated, or needs intervention. The Queen reads the room—detecting patterns humans miss. Invoke when: progress stalled, energy feels off, team might be looping.
Agent Claude Code
Use for running test commands (pytest, npm test, etc.) and reading test files via Bash, Read, and Grep. Performs no source edits and no git side effects. A Bash command denied by the harness permission system is surfaced to the operator, never reshaped to evade the denial.
Agent Claude Code
Use for quarantined web research — fetching URLs, running web searches, and reading local notes to produce structured findings. Returns JSON results only; never edits files or executes shell commands. Requires WI-8 (devcontainer) to be merged before being safe to dispatch in production.
Command Claude Code
Systematically address PR review comments. Fetches all threads, categorizes by status (acknowledged, silently fixed, unaddressed), and guides user through posting replies and implementing fixes with explicit approval.
Command Claude Code
Advanced Code Review Phase 2: Context Analysis - load previous reviews, PR history, declined items.
Command Claude Code
Advanced Code Review Phase 1: Strategic Planning - scope analysis, risk categorization, priority ordering.
Command Claude Code
Advanced Code Review Phase 5: Report Generation - produce final deliverables including Markdown report and JSON summary.
Command Claude Code
Advanced Code Review Phase 3: Deep Review - multi-pass code analysis, finding generation with severity classification.
Command Claude Code
Advanced Code Review Phase 4: Verification - fact-check findings against codebase, remove false positives.
Command Claude Code
Audit test suites for Green Mirage anti-patterns: tests that pass but don't verify behavior.
Command Claude Code
Phases 2-3 and 7 of auditing-green-mirage: systematic line-by-line audit, the Green Mirage Patterns, named assertion shapes, and the fix-verification Test Adversary prompt.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: