vectorspacelab/arex-skill/accuracy-pair-protocol

Claude Code and agent skill by vectorspacelab/arex-skill

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Validate paired ID/OOD accuracy records before fitting accuracy-on-the-line calibration experiments.

Install

npx vskill@latest install vectorspacelab/arex-skill/accuracy-pair-protocol

SKILL.md

The same SKILL.md was published earlier as vectorspacelab/arex-skill/accuracy-pair-protocol.

Accuracy Pair Protocol

Use this skill when a recovery or benchmark script needs to build the paired in-distribution and out-of-distribution accuracy panel used by Accuracy on the Line. Do not use it to fit the line or decide whether the proxy passes; it only owns record validation and deterministic ordering.

Inputs

  • A JSON list of objects with model_id, id_accuracy, and ood_accuracy.
  • Optional metadata fields such as architecture, training duration, or shift name.

Outputs

  • A JSON object containing validated records, count, id_range, ood_range, and provenance.
  • Validation errors for missing fields, duplicate model ids, nonnumeric accuracies, or values outside [0, 1].

Workflow

  1. Parse records from JSON or receive them as Python dictionaries.
  2. Require one paired ID and OOD accuracy for every model id.
  3. Enforce numeric ranges and unique model ids.
  4. Sort records by model_id so downstream line fitting is reproducible.
  5. Preserve optional metadata without adding predictions or pass/fail decisions.

Validation

Run python scripts/accuracy_pairs.py --input tests/fixtures/valid_pairs.json or validate the tree with the Distiller validate_skill_tree.py --run-tests command.

Limitations

This skill does not download datasets, train models, fit the calibration line, or evaluate a recovery target. Synthetic records must be labeled by the caller as proxy evidence.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/vectorspacelab/arex-skill/accuracy-pair-protocol/badge)](https://verified-skill.com/skills/vectorspacelab/arex-skill/accuracy-pair-protocol)

Something wrong with this skill or its page? Tell us.