Multi-agent code review using Agent Teams. Summons up to 7 built-in Ashes in standard mode, each with their own dedicated context window. Handles scope selection, team creation, review orchestration, aggregation, verification, and cleanup. Optional `--deep` runs multi-wave deep review with up to 18 Ashes across 3 waves. Phase 1.5 adds UX reviewers when frontend files are detected (UX subsystem hardcoded in v3.x). Phase 1.7 adds data flow integrity reviewer (FLOW prefix) when 2+ stack layers detected in diff.
Full plan-to-merged-PR pipeline with checkpoint framework, QA phases, and multi-agent orchestration. Use when running end-to-end from plan to merge, resuming an interrupted arc with --resume after a crash or session end, or when any named phase fails (forge, work, code-review, mend, test, ship, merge, etc. — see body for full phase list). Keywords: arc, pipeline, --resume, checkpoint, convergence, forge, mend, ship, merge.
Full codebase audit using Agent Teams. Sets scope=full and depth=deep (by default), then delegates to the shared Roundtable Circle orchestration phases. Summons up to 7 built-in Ashes (custom Ashes wired in orchestration layer in v3.x). Optional `--deep` runs multi-wave investigation with deep Ashes. Phase 0.45 context building spawns context-builder agent to map trust boundaries, invariants, and state flows before vulnerability hunting (auto for deep). Supports `--focus` for targeted audits. Supports `--incremental` for stateful, prioritized batch auditing with 3-tier coverage tracking (file, workflow, API) and session-persistent audit history.
Collaborative exploration of features and ideas through structured dialogue. Explore WHAT to build before planning HOW. Three modes: Solo (conversation), Roundtable (agent advisors engage user), Deep (advisors + elicitation sages). Produces persistent brainstorm documents in docs/brainstorms/. Use when: "brainstorm", "explore idea", "what should we build", "discuss feature", "thao luan", "kham pha y tuong", "brainstorm this", "let's think about".
ACH-based parallel debugging. Spawns multiple hypothesis-investigator agents to investigate competing hypotheses simultaneously. Use when bugs are complex, when single-agent debugging hits 3+ failures, or when root cause is unclear. Also triggers when test-failure-analyst returns LOW confidence during arc Phase 7.7. Keywords: debug, investigate, hypothesis, root cause, parallel debugging, ACH, competing hypotheses, falsify, evidence, multi-agent debug.
Multi-agent planning workflow using Agent Teams. Combines brainstorm, research, validation, synthesis, shatter assessment, forge enrichment, and review into a single orchestrated pipeline with dependency-aware task scheduling.
Deepen an existing plan with Forge Gaze topic-aware enrichment. Summons specialized Ashes to enrich each section with expert perspectives. Can target a specific plan or auto-detect the most recent one.
Use when assessing blast radius of changes, when understanding WHY code was written a certain way, when planning risky modifications, or when a refactor could have hidden side effects. Traces WHAT changes, WHY it exists (git archaeology), and HOW RISKY the area is (churn + ownership metrics). Keywords: impact analysis, blast radius, risk assessment, code archaeology, why was this written.