Skip to content
Get packs

Paperclip Evals Navigator

Guides an agent through choosing, validating, and reporting Paperclip evaluation runs with evidence preserved.

Source Download skill
Repository
paperclipai/paperclip
Stars
110,078
Licence
MIT
Files
1 file

What it does

This skill teaches an AI agent how to work with the evaluation system in the Paperclip repository. It tells the agent which of two evaluation families fits a request, how to find the right repository checkout, how to run safe credential-free checks first, and how to report results with their source revisions, cost, and failure causes attached. It is a set of working rules and pointers to other project docs rather than a tool that runs evaluations on its own.

Great for

  • Engineers maintaining the Paperclip runner or its end-to-end test suite who want an agent to interpret eval results consistently
  • Contributors who need to decide whether a change belongs in a Runner Eval or a Product E2E eval
  • Reviewers who ask an agent to summarise a live eval run with its provenance and failure classification

Skip it if

  • Users outside the Paperclip project, since it points to repository-specific files and private sibling checkouts that a general user will not have
  • Launching paid or live provider runs without explicit authorization, which the skill itself forbids without user approval and configured credentials

Try asking

  • “Which evaluation family should I use to test a new runner protocol behaviour?”
  • “Validate the Product E2E runner setup without spending credits”
  • “Summarise the latest eval history and explain whether the failures are product or infrastructure problems”
  • “Refresh the public eval viewer from retained evidence without making new provider calls”
Step-by-step guide: how to use Paperclip Evals NavigatorInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    paperclip-evals.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip paperclip-evals.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip paperclip-evals.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Ask for the credential-free validation commands first; only request a live run once you have confirmed credentials and authorization.
  • Point the agent at the repository explicitly or set PAPERCLIP_EVALS_ROOT so it does not guess paths.
  • Check that reported results cite the source revision and grader version, since the skill requires them for every interpretation.

What's inside

  • SKILL.md

Redistributed unmodified from paperclipai/paperclip under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.