Self Eval

Self-Eval: Honest Work Evaluation

ultrathink

Tier: STANDARD Category: Engineering / Quality Dependencies: None (prompt-only, no external tools required)

Description

Self-eval is a Claude Code skill that produces honest, calibrated work evaluations. It replaces the default AI tendency to rate everything 4/5 with a structured two-axis scoring system, mandatory devil's advocate reasoning, and cross-session anti-inflation detection.

The core insight: AI self-assessment converges to "everything is a 4" because a single-axis score conflates task difficulty with execution quality. Self-eval separates these axes, then combines them via a fixed matrix that the model cannot override.

Features

• Two-axis scoring — Independently rates task ambition (Low/Medium/High) and execution quality (Poor/Adequate/Strong), then combines via a lookup matrix
• Mandatory devil's advocate — Before finalizing, must argue for both higher AND lower scores, then resolve the tension
• Score persistence — Appends scores to .self-eval-scores.jsonl in the working directory, building history across sessions
• Anti-inflation detection — Reads past scores and flags clustering (4+ of last 5 identical)
• Matrix-locked scoring — The composite score comes from the matrix, not from direct selection. Low ambition caps at 2/5 regardless of execution quality

Usage

After completing work in a Claude Code session:

/self-eval

With context about what to evaluate:

/self-eval the authentication refactor we just completed

The skill reviews the conversation history, identifies what was accomplished, scores it on both axes, runs the devil's advocate check, and appends the result to the score history file.

Examples

Example 1: Feature Implementation

/self-eval added pagination to the user list API

Output:

## Self-Evaluation

**Task:** Added cursor-based pagination to the user list API endpoint
**Ambition:** Medium — New feature with established patterns, but required careful backwards compatibility
**Execution:** Strong — All tests pass, handles edge cases, existing clients unaffected

**Devil's Advocate:**
- Lower: Pagination is well-documented. The pattern was copied from another endpoint with minor changes. Could argue Low ambition.
- Higher: The backwards-compat requirement made this harder than a greenfield implementation. Had to support both offset and cursor modes.
- Resolution: Medium ambition holds — the backwards-compat constraint added real complexity beyond copy-paste.

**Score: 4** — Solid execution on meaningful work with real constraints.

Example 2: Bug Fix

/self-eval

Output:

## Self-Evaluation

**Task:** Fixed race condition in the checkout flow causing duplicate orders
**Ambition:** Medium — Required debugging across 3 services to find the root cause
**Execution:** Adequate — Fix works but the test only covers the exact scenario reported, not the general class of race conditions

**Devil's Advocate:**
- Lower: The fix was a single line (adding a mutex). Once found, it was trivial.
- Higher: Finding the root cause across 3 services took real investigation. The mutex placement required understanding the full flow.
- Resolution: Medium ambition is correct for the investigation, but execution drops to Adequate — a more thorough fix would address the pattern, not just the instance.

**Score: 3** — Good debugging work but the fix is narrow.

What to Evaluate

$ARGUMENTS

If no arguments provided, review the full conversation history to identify what was accomplished this session. Summarize the work in one sentence before scoring.

How to Score — Two-Axis Model

Score on two independent axes, then combine using the matrix. Do NOT pick a number first and rationalize it — rate each axis separately, then read the matrix.

Axis 1: Task Ambition (what was attempted)

Rate the difficulty and risk of what was worked on. NOT how well it was done.

• Low (1) — Safe, familiar, routine. No real risk of failure. Examples: minor config changes, simple refactors, copy-paste with small modifications, tasks you were confident you'd complete before starting.
• Medium (2) — Meaningful work with novelty or challenge. Partial failure was possible. Examples: new feature implementation, integrating an unfamiliar API, architectural changes, debugging a tricky issue.
• High (3) — Ambitious, unfamiliar, or high-stakes. Real risk of complete failure. Examples: building something from scratch in an unfamiliar domain, complex system redesign, performance-critical optimization, shipping to production under pressure.

Self-check: If you were confident of success before starting, ambition is Low or Medium, not High.

Axis 2: Execution Quality (how well it was done)

Rate the quality of the actual output, independent of how ambitious the task was.

• Poor (1) — Major failures, incomplete, wrong output, or abandoned mid-task. The deliverable doesn't meet its own stated criteria.
• Adequate (2) — Completed but with gaps, shortcuts, or missing rigor. Did the thing but left obvious improvements on the table.
• Strong (3) — Well-executed, thorough, quality output. No obvious improvements left undone given the scope.