Agent-Skills.md

Agent Skills: Review Against SOLID Principles — Single-Pass Baseline (Arm A)

Ensemble SOLID review — fans the ruleset across overlapping focused workers over the same code, then reduces by corroboration.

UncategorizedID: Jamie-BitFlight/claude_skills/review-against-solid-principles

Repository

Jamie-BitFlight/claude_skills
Jamie-BitFlightLicense: MIT
6110

Install this agent skill to your local

pnpm dlx add-skill https://github.com/Jamie-BitFlight/claude_skills/tree/HEAD/examples/solid-review-ab/single-high-intelligence-agent/.claude/skills/review-against-solid-principles

Skill Files

Browse the full folder contents for review-against-solid-principles.

Download Skill

Loading file tree…

examples/solid-review-ab/single-high-intelligence-agent/.claude/skills/review-against-solid-principles/SKILL.md

Skill Metadata

Name
review-against-solid-principles
Description
"Single-pass SOLID review — dispatches one reviewer over the complete ruleset, emitting the fixed candidate schema."

Review Against SOLID Principles — Single-Pass Baseline (Arm A)

One agent holds the entire SOLID ruleset and reviews the target in one pass.

Procedure

Fixed paths (relative to this arm directory, the claude -p working directory):

  • Ruleset: ../ruleset/solid-rules.json
  • Targets: every *.py under ../corpus/cases/
  • Output: ./findings/findings.md
  1. Dispatch ONE principles-reviewer subagent. Give it the ruleset path, the list of target files, and the output path. It reads the full ruleset and every target, applies all rules in a single pass, and writes the fixed-schema findings to ./findings/findings.md.
  2. Do NOT partition the rules, fan out, or spawn more than one reviewer. One subagent does the entire review.
  3. Write the findings to ./findings/findings.md.

The principles-reviewer agent emits the fixed candidate schema.

Review Against SOLID Principles — Single-Pass Baseline (Arm A) Skill | Agent Skills