Skills
.Work. You
Rest

collab-evals

Community

Run collab evals and capture manifest evidence.

Data & Analytics#evaluation#multi-agent#evidence#manifest#RLM#pause-resume#collab
AuthorKbediako
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Collab-evals provide a framework to run repeatable multi-agent evaluation scenarios (symbolic RLM, large-context interactions) and to preserve evidence via manifest-backed outputs, reducing ad-hoc experimentation and enabling audit trails.

Core Features & Use Cases

  • Orchestrates collab-driven evaluations across multi-agent workflows including symbolic RLM and large-context tests.
  • Supports pause/resume, long-running experiments, and checkpointing for resilience.
  • Generates manifest-backed evidence and updates documentation with findings for traceability and reproducibility.

Quick Start

  1. Pick the scenario(s) for evaluation:
  • Large-context symbolic RLM with collab subcalls.
  • Multi-hour refactor with checkpoints.
  • 24h pause/resume context-rot regression.
  • Multi-day initiative (48–72h) with multiple resumes.
  1. Ensure task context:
  • export MCP_RUNNER_TASK_ID=<task-id>
  1. Run the scenario using codex-orchestrator start <pipeline> --format json and record the manifest path.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: collab-evals
Download link: https://github.com/Kbediako/CO/archive/main.zip#collab-evals

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 223,000+ vetted skills library on demand.