collab-vs
The agent brainstorms with a partner LLM in alternating turns, building on each other's ideas. Synthesizes the best ideas into a plan.
collab
Multiple LLMs collaboratively brainstorm solutions, building on each other's ideas across rounds. Agent synthesizes the best ideas into a plan.
consult-llm
How to invoke the consult-llm CLI. Canonical reference for the invocation contract, flags, stdin/stdout format, and multi-turn. Load this before calling consult-llm from any workflow skill (/consult, /collab, /debate, /collab-vs, /debate-vs).
consult
Consult an external LLM with the user's query.
debate-vs
The agent debates an opponent LLM through a multi-turn conversation, then synthesizes the best approach and implements.
debate
LLMs propose and critique approaches, agent moderates the debate and synthesizes the best solution, then implements.
implement
Explicit preset-driven implementation workflow. Use only when the user invokes `/implement` or another skill explicitly delegates to it. Do not trigger for ordinary coding requests, straightforward follow-up edits, fixes with an established design, or requests to amend an existing commit.
panel
Role-specialized LLM panel analyzes a task from asymmetric expert lenses (architect, security, maintainability, test-strategist by default). Agent synthesizes a trade-off resolution.
phased-implement
Coordinator workflow for multi-phase implementation across workmux worktrees. Generates or loads a master plan, dispatches phase agents using presets, verifies sentinels, merges serially, and performs integration verification.
review-panel
Standalone multi-model code review of an existing diff. Multiple LLMs review in parallel; agent deduplicates, prioritizes by severity/confidence, and optionally applies localized fixes.
review
Collect critical feedback from all registered LLMs on an artifact (architecture doc, implementation, plan). Intellectual debate with push-back — no sycophancy. Reports findings and unresolved disagreements.
workshop
Interactive design session — agent facilitates a clarifying dialogue with the user, fans out to multiple LLMs in parallel for divergent approach generation, lets the user pick one, then co-designs the chosen approach with optional multi-LLM critique before saving.