How do I install the Eval Outcomes skill?

Run npx skillmds@latest add majiayu000/eval-outcomes in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the Eval Outcomes skill do?

Holdout-safe Outcomes grading: project the locked eval substrate, then ingest one verdict. It is listed under Coding & Dev Tools on SkillMD.

Is Eval Outcomes safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: CAUTION, Skill Scanner: PASS. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Eval Outcomes?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Eval Outcomes free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Eval Outcomes?

majiayu000 (@majiayu000) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Eval Outcomes by majiayu000 — Agent Skill

Name: Eval Outcomes
Rating: 4.2 (5 reviews)
Author: majiayu000

File contents

$eval-outcomes — Outcomes as a Projection of the Locked Eval Substrate (Codex Native)

Quick Ref: Grade an agent's output via Outcomes (or any model) without forking the bar. ao eval outcomes compile projects a locked Task into a holdout-safe rubric (refuses to leak target/ground_truth — Managed Agents are not ZDR); grade it; ao eval outcomes ingest writes the score back as the one council verdict record. Outcomes is a projection, never an alternate authority.

Codex path

Codex has no Managed Agents loop. Call the same ao eval outcomes compile <input.json> to get a holdout-safe rubric, grade it locally (Inspect AI over the dev split, or the bushido llama.cpp Qwen grader over tailnet), then ao eval outcomes ingest <score.json> --json. Net: Codex never touches the cloud Outcomes API but produces a byte-identical verdict record.

Instructions

Load and follow the skill instructions from the sibling SKILL.md — OR read skills/eval-outcomes/SKILL.md in the host repo for the canonical specification. Then apply the three-phase Workflow (compile → grade → ingest), honoring the Critical Constraints (never send holdout target/ground_truth/PII; carry judge_content_hash; register a global Dolt burn for holdout grades).

Frequently asked questions

Eval Outcomes