Index of every page on this site: llms.txt
All skills
neondatabase avatar

/score-eval

@d3601d9 official
Run it yourself

Agent Skills for Neon Severless Postgres

  • 1 file
  • 517 B
  • Updated 8 months ago
  • GitHub
Use this Skill: https://skilld.dev/gh/neondatabase/agent-skills/score-eval

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈13 tokens for metadata: the name and description. ≈115 when used: this file.

SKILL.md has missing or invalid fields
  • The description in the SKILL.md header is missing or is not text.
Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Score the eval diff at $ARGUMENTS against the eval rubric.

  1. Read the diff file at the path provided
  2. Read the eval rubric at eval-rubric.md
  3. Read the original fixture app in fixtures/hono-drizzle-app/ for comparison
  4. For each problem P1-P5, answer the Detected? and Fixed? questions from the rubric as yes or no
  5. Append a row to results.csv — fill in all fields you can determine from the diff and context. Leave fields you can't determine empty.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at d3601d9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 8 months ago
disable-model-invocation
true
  • Testing
  • eval
  • rubric
  • scoring
  • diff
  • csv
  • automation
  • qa

README badge

README badge for neondatabase/agent-skills/score-eval

Reads a diff file and evaluates it against a rubric defined in eval-rubric.md, answering yes/no questions for five problem categories (P1–P5) and appending results to a CSV. Designed for grading code changes in a Hono + Drizzle fixture app, not for general-purpose code review.

Generated from the current SKILL.md.

What does this skill evaluate?
It scores code diffs against a predefined eval rubric with five problem categories (P1-P5), determining whether each problem was detected and fixed in the diff.
What files does this skill require?
The skill requires a diff file (passed as an argument), an eval-rubric.md file defining the scoring criteria, and a fixtures/hono-drizzle-app/ directory containing the original app for comparison.
What does this skill output?
It appends a row to results.csv with fields populated based on whether each problem was detected and fixed, leaving fields empty if they cannot be determined from the diff.
Can an AI model invoke this skill directly?
No. The skill has disable-model-invocation set to true, meaning it must be called explicitly by the user or another process, not automatically by Claude.

Generated from the current SKILL.md. These answers refresh after source changes.