Coordinate multiple CLI agents to generate, anonymize, and merge implementation plans.
npx skills add https://github.com/am-will/codex-skills --skill llm-council-am-will
Please help me install this Agent Skill. Skill: llm-council Source: https://github.com/am-will/codex-skills/tree/main/skills/llm-council Command: npx skills add https://github.com/am-will/codex-skills --skill llm-council-am-will
The LLM Council skill coordinates a configurable, multi-member CLI planning workflow to produce independent implementation plans, anonymize outputs, and merge them into a single final plan for auditability.
Run the council on a task spec:
High-intent search queries and answers about installing and using this skill.
Multi-agent planning coordination is handled by orchestrating independent CLI agents like Codex and Claude to generate drafts. The system anonymizes, shuffles, and judges these drafts to synthesize a unified, documented final plan with metadata.
Bias-resistant planning prevents model-specific bias by anonymizing and shuffling independently generated implementation drafts. A judge agent evaluates the content blindly, synthesizing a final plan based on merit rather than the originating CLI agent.
Run the council workflow using python3 scripts/llm_council.py run --spec path/to/spec.json. Optional flags include --timeout, --config, and --out to customize execution parameters and capture the final output.
Yes, the planning council applies to software projects and broader initiatives requiring bias-resistant planning, auditability, and structured outputs. It enforces task_spec validation and produces Markdown outputs with preserved run artifacts for traceability.
The workflow depends on available CLI agents like Codex, Claude, Gemini, and OpenCode, alongside Python3 for script execution. It requires a valid task spec JSON file to validate and initiate the parallel plan generation process.
Run artifacts are preserved under llm-council/runs to ensure auditability and traceability of the multi-agent planning process. This allows users to review anonymized drafts, judge decisions, and final plan metadata post-execution.
ClementATHSpawn multiple CLI workers in tmux panes to execute tasks in parallel.
SpiralCloudOmegaAutomates anonymous multi-agent code-artifact review with consensus scoring and remediation guidance.
lucking7Execute structured work plans with inline, serial, or parallel subagents.
0xharryriddleCoordinate multiple CLI planning agents to generate, anonymize, and merge implementation plans.
