paseo-committee

Convenes two contrasting high-reasoning agents via Paseo to analyze problems and converge on a plan.

Updated Nov 25, 2024
One-click install
npx skills add https://github.com/cbrostrom/dotfiles --skill paseo-committee-cbrostrom
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: paseo-committee
Source: https://github.com/cbrostrom/dotfiles/tree/main/stow/agents/.agents/skills/paseo-committee
Command: npx skills add https://github.com/cbrostrom/dotfiles --skill paseo-committee-cbrostrom

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When you are stuck in a loop, tunnel-visioning on a wrong approach, or facing a hard planning problem, a single agent's perspective often is not enough. This Skill forms a committee of two high-reasoning agents with contrasting profiles to perform root cause analysis and produce a converged plan. ## Core Features & Use Cases - Contrasting Agent Selection: Picks two committee members from different provider families using configured Paseo agent profiles, maximizing diversity of reasoning styles. - Parallel Analysis with Convergence: Both agents receive the same problem-level prompt in parallel, then disagreements are resolved by passing their arguments between each other until they converge. - Analysis-Only Safety: Every prompt enforces a strict no-edits rule, so committee members never modify files or write code. - Use Case: You have been debugging a flaky deployment for hours with no progress. Invoke the committee to get two independent root-cause analyses, reconcile their disagreements, and receive a consensus plan. ## Quick Start Ask the agent to form a committee to analyze why the build keeps failing and produce a plan to fix it.

Frequently Asked Questions about paseo-committee

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get a second opinion from another AI agent when stuck on a problem?▼

Invoke the committee workflow, which creates two high-reasoning agents in parallel via Paseo with the same problem-level prompt. Their disagreements are resolved by passing arguments between them until they converge on a plan.

How are committee members selected in Paseo?▼

Members are chosen from configured agent profiles after calling list_profiles, picking one planning or root-cause-oriented profile and one contrasting high-reasoning profile from a different provider family. If fewer than two suitable profiles exist, Paseo's provider discovery fallback fills the gap.

Can committee agents edit or create files during analysis?▼

No. Every prompt sent to a committee member ends with a strict no-edits suffix forbidding file creation, modification, deletion, or code writing. The committee is analysis-only by design.

Why does the committee take a long time to respond?▼

High-reasoning models can think for 15 to 30 minutes per response. The workflow relies on Paseo's finish notification rather than polling, so long waits are expected and should not be interrupted.

When should I not use a multi-agent committee?▼

Avoid it for routine tasks with clear solutions, since running two high-reasoning agents adds latency and cost. It is intended for being stuck, looping, or facing genuinely hard planning problems where a single perspective has failed.