Skills
skills/lanyasheng/auto-improvement-orchestrator-skill/skill-forge/Gen Agent Trust Hub

skill-forge

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/forge.py utilizes subprocess.run to execute specific Python scripts for evaluation and orchestration. These scripts are located in predictable paths within the user's home directory (~/.claude/skills/) and are part of the intended integration between development tools.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it extracts and reuses content from external SKILL.md and skill_spec.yaml files. \n
  • Ingestion points: Content is ingested from user-provided file paths in scripts/task_suite_generator.py and scripts/forge.py.\n
  • Boundary markers: The output task_suite.yaml does not employ delimiters to isolate extracted text within generated prompts or rubrics.\n
  • Capability inventory: The skill performs file system writes and executes specific local scripts via subprocess.\n
  • Sanitization: Extracted text is used without robust sanitization, relying on regex for basic structure extraction rather than security filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 03:25 AM