Skip to main content
skill CX Codex

llm-council · rohitg00 — Claude Code skill: what it does, how to install

llm-council runs multi-model deliberation in three stages: parallel drafts, anonymous peer rankings, and final synthesis by a designated chairman.

MIT ★ 2,966 updated 2026-09-29

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflow
Open on GitHub

What llm-council does

The skill implements Andrej Karpathy's LLM Council pattern for high-stakes plan critiques and architectural debates. Work proceeds through three mandatory phases: models draft answers in parallel, rank each other's anonymized outputs, and a designated chairman produces the final consensus.

Deliberation artifacts persist to ~/.pro-workflow/council/<session-id>/. The directory stores verbatim model outputs, ranking matrices, and a human-readable markdown summary. Passing --wiki <slug> writes the transcript directly into a wiki page indexed by SQLite FTS5.

Deliberation cost grows quadratically with council size because every model evaluates every peer. Running the script requires Node.js and explicit provider keys like PRO_WORKFLOW_ANTHROPIC_API_KEY or an MCP provider runner.

What it helps with

  • evaluate architectural trade-offs using multiple model perspectives
  • review high-risk plans through anonymous peer ranking
  • resolve conflicting project rules by voting across providers
  • archive multi-LLM debate transcripts to a searchable wiki

Example prompts for the agent

  • Run a council on whether we should adopt event sourcing
  • /council "Should we migrate from REST to tRPC?"
  • /wiki council agent-memory "Should we adopt episodic memory in our agents?"

How to install llm-council

1Plugin marketplace

/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflow

This installs the whole pro-workflow plugin with all its skills

2Skills CLI (Vercel)

npx skills add rohitg00/pro-workflow --skill llm-council -a claude-code

3Manually: put the files in place

~/.claude/skills/llm-council/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/llm-council/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about llm-council

How to install llm-council in Claude Code?

Add the marketplace and install the package by running /plugin marketplace add rohitg00/pro-workflow followed by /plugin install pro-workflow@pro-workflow. Alternatively, run npx skills add rohitg00/pro-workflow --skill llm-council -a claude-code.

What does llm-council do?

It coordinates several LLMs to answer a query independently, lets them rank peer answers anonymously, and has a chairman model synthesize the conclusions.

Which providers are supported?

It supports Anthropic, OpenAI, OpenRouter, Fireworks, and custom endpoints using OpenAI-compatible APIs configured via environment variables.

Source text

SKILL.md · MIT
---
name: llm-council
description: Provider-agnostic multi-LLM deliberation. Three phases — independent responses, cross-model anonymized ranking, chairman synthesis. Credentials come from explicit plugin configuration or pro-workflow-specific CLI variables. Persists transcript to a wiki page when --wiki <slug> is passed. Use when the user wants multiple AI perspectives, consensus-building, or the "LLM Council" approach for high-stakes reviews, plan critique, or contested learning rules.
user-invocable: true
---

# LLM Council

Karpathy's LLM Council pattern with Anthropic and OpenAI-compatible providers.

## When to use

- High-stakes plan review (`/plan` crosses N-file threshold)
- Conflicting learning-rules → re-resolve via vote
- User invokes `/council "<query>"` or `/wiki council`
- Architecture decisions where you want multiple viewpoints captured
- Persisting deliberation as a wiki page for future reference

## Three phases

1. **Independent**: each model answers in parallel
2. **Ranking**: each model ranks anonymized peer responses
3. **Synthesis**: chairman model reads all responses + rankings → final answer

## Provider config

Configure optional keys in the plugin configuration dialog. Use the `providers` MCP server's `run_provider_task` tool with `task: "council"` and `args: ["run", "<query>", "--provider", "openai"]`. For provider status, pass `args: ["providers"]`. Never request keys in chat or read them from existing credentials or shell configuration.

For standalone CLI use, explicitly set a pro-workflow-specific variable below. The first configured provider is the default; `--provider` selects one explicitly. See [provider configuration](../../references/provider-configuration.md).

| Standalone CLI variable | Provider | Default base URL |
|---------|----------|------------------|
| `PRO_WORKFLOW_ANTHROPIC_API_KEY` | Anthropic | `https://api.anthropic.com` |
| `PRO_WORKFLOW_OPENAI_API_KEY` | OpenAI | `https://api.openai.com/v1` |
| `PRO_WORKFLOW_OPENROUTER_API_KEY` | OpenRouter | `https://openrouter.ai/api/v1` |
| `PRO_WORKFLOW_FIREWORKS_API_KEY` | Fireworks | `https://api.fireworks.ai/inference/v1` |
| `LLM_COUNCIL_BASE_URL` + `PRO_WORKFLOW_LLM_COUNCIL_API_KEY` | Custom OpenAI-compat | (user-supplied) |

Override per-run with `--provider openai|anthropic|openrouter|fireworks|custom`.

Default model rosters per provider live in `scripts/council.js` and can be overridden via `--models` CSV and `--chairman <id>`.

## Commands

In a plugin session, pass these runner arguments through `run_provider_task`. The direct commands below are for standalone CLI installations with explicit credentials.

```
node $SKILL_ROOT/scripts/council.js run "<query>" [--models id1,id2,id3] [--chairman id] [--provider <name>] [--wiki <slug>]
node $SKILL_ROOT/scripts/council.js providers
node $SKILL_ROOT/scripts/council.js show <session-id>
```

`--wiki <slug>` writes the full transcript to `<wiki>/derived/council/<session-id>.md` and registers it via `wiki-cli.js page` so it shows in FTS5 search.

## Output

Each session writes:

```
~/.pro-workflow/council/<session-id>/
├── config.json           # query, models, chairman, provider
├── phase1_responses.json # raw API responses per model
├── phase2_rankings.json  # anonymized ranking outputs
├── phase3_synthesis.txt  # chairman's final answer
└── final_output.md       # human-readable bundle
```

Console prints the markdown bundle. Pipe to `pbcopy` / `tee` as needed.

## Hard rules

1. Never skip the ranking phase. It's the core of the council pattern.
2. Save raw responses to disk verbatim. No summarization in storage.
3. Anonymize responses for ranking — models see `Response A/B/C/...`, not peer names.
4. The chairman sees both real names AND rankings.
5. Display all three phases to the user. No phase elision.

## Cost awareness

The script logs per-call latency + tokens on supported providers. Multiply by your provider rate to estimate. Council cost grows linearly with `len(models)^2` (each model ranks all others) plus the chairman.

Default council size: 3-5 models. More models = exponentially more ranking calls.

## Use with wiki

```
/wiki council agent-memory "should we adopt episodic memory in our agents?"
```

Loads `agent-memory` wiki context as system prompt prefix, runs council, persists transcript as `wiki/derived/council/<id>.md`. The transcript becomes searchable via `/wiki ask`.

© Rohit Ghumare · MIT · rohitg00/pro-workflow

Similar skills

All skills for Agent setup: memory, context, orchestration →

llm-council by other authors comes first.

llm-council aiwithremy llm-council is a Claude Code skill that launches five advisor sub-agents with distinct thinking styles to analyze a question independently, review each other's answers, and produce a final verdict. CX Codex +1 1,600installs llm-council am-will The llm-council skill runs several CLI agents as independent planners, strips their identities, and merges the results into one final plan. Useful when a single model's bias is the problem. CX Codex +1 1,200installs llm-council happycapy-ai llm-council is a Claude Code skill that sends one prompt to several AI models and displays their replies side by side, with shared-answer synthesis and anonymous voting. CX Codex +1 ★ 138repo stars LLM Council shuntacurosu LLM Council coordinates multiple LLMs across three stages: independent responses, anonymized peer grading, and chairman synthesis. CX Codex +1 5installs Tokonomix Council TokonoMix Tokonomix Council runs one question past several models, then has an outside judge turn their findings into a single answer. Useful when a call is too costly to leave to one model. CX Codex +1 ★ 1repo stars council boshu2 council — a skill for Claude Code gathers independent views and synthesizes their agreements and disagreements. Use it to compare plans, discuss options, or summarize several reviews. CX Codex +1 3,100installs council-decision mylukin The council-decision skill gathers opinions from Claude, Codex, and Gemini on a complex question, has each model evaluate the others' answers, then synthesizes a final recommendation. Use it when architectural choices are unclear. CX Codex +1 4installs hive-mind-advanced ruvnet hive-mind-advanced is a Claude Code skill that coordinates specialized agents through a queen coordinator, consensus, and shared memory. It distributes tasks across complex projects. CX Codex +1 1,100installs tribunal a-tokyo tribunal coordinates a multi-agent review loop with isolated verifiers and an adversary to gate deliverable releases. Stops hidden failure modes on critical tasks before deployment. CX Codex +1 891installs

From the same set83

The whole set →
wiki-builder wiki-builder is a Claude Code skill that creates a structured knowledge base from Markdown pages and an FTS5 index in SQLite. It persists between sessions and is queryable by the agent. CX Codex +1 ★ 2,966repo stars token-efficiency token-efficiency cuts conversational fluff, caps tool executions per task, and stops Claude Code from repeatedly editing working implementations. CX Codex +1 ★ 2,966repo stars wiki wiki is a Claude Code command that builds and maintains persistent research wikis. It stores markdown files, indexes them in SQLite FTS5, and retrieves context via BM25. CX Codex +1 ★ 2,966repo stars wiki-viewer wiki-viewer writes a pro-workflow wiki out as a single HTML file with pages, sources, claims, a seed queue, a link graph and search. Useful for handing research state to someone else. CX Codex +1 ★ 2,966repo stars design-engineering The design-engineering skill guides interface animation choices and UI reviews. Use it when refining a component, transition, modal, or other interactive state. CX Codex +1 ★ 2,966repo stars permission-tuner The permission-tuner skill reads permission prompts and auto mode denials from your session logs, then suggests permission rules and autoMode entries to reduce interruptions. CX Codex +1 ★ 2,966repo stars batch-orchestration batch-orchestration decomposes large codebase changes into independent units and runs parallel agents in isolated git worktrees. CX Codex +1 ★ 2,966repo stars insights insights turns your pro-workflow history into numbers: edits per session, rules you actually apply, mistakes that keep coming back. Useful when you want to see where the agent keeps slipping. CX Codex +1 ★ 2,966repo stars

Project sponsors

Companies that keep this analytics open