claude-code-headless

Run Claude Code programmatically without interactive UI. Triggers on: headless, CLI automation, --print, output-format, stream-json, CI/CD, scripting.

Invalid

This skill can't be scored yet

Validation errors are blocking scoring. Review and fix them to unlock Quality, Impact and Security scores. See what needs fixing →

Evaluation results

92%

14%

Automated Code Review in CI Pipeline

CI/CD integration with error handling

Criteria

Without context

With context

Headless flag used

100%

JSON output format

100%

Raw JSON saved

100%

Result text extracted

100%

System prompt appended

40%

100%

Permission mode set

is_error field checked

100%

Non-zero exit on error

100%

Error message to stderr

50%

100%

Tool restriction

50%

100%

Without context: $0.4144 · 1m 50s · 22 turns · 74 in / 6,058 out tokens

With context: $0.3934 · 1m 20s · 21 turns · 456 in / 4,857 out tokens

93%

-7%

Automated Three-Stage Codebase Analysis

Multi-turn session management

Criteria

Without context

With context

Headless flag used

100%

37%

JSON output format

100%

session_id captured

100%

Resume flag used

100%

JSON responses saved

100%

Text results extracted

100%

Three distinct stages

100%

Session documented

100%

75%

Progress logged

100%

No hardcoded session IDs

100%

Without context: $0.3678 · 1m 38s · 18 turns · 55 in / 5,588 out tokens

With context: $0.3941 · 1m 20s · 23 turns · 69 in / 4,609 out tokens

100%

15%

Read-Only Dependency Audit Tool

Tool restriction and stream output

Criteria

Without context

With context

Headless flag used

100%

Stream-JSON format

100%

Streamed output saved

100%

allowedTools restricted

100%

disallowedTools set

100%

Bypass permissions

100%

Verbose flag included

100%

Restrictions documented

100%

Directory argument used

100%

Without context: $0.3561 · 1m 40s · 21 turns · 28 in / 5,296 out tokens

With context: $0.3643 · 1m 30s · 22 turns · 310 in / 4,601 out tokens

Repository: NeverSight/skills_feed
Commit: 5342bca

Evaluated: 13 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Automated Code Review in CI Pipeline Automated Three-Stage Codebase Analysis Read-Only Dependency Audit Tool

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.