skills/tercel/code-forge/impl/Gen Agent Trust Hub

impl

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it reads task definitions and state configurations from the local filesystem to populate sub-agent prompts. This ingestion is essential for its primary function as a task coordinator.\n
  • Ingestion points: Reads state.json and task markdown files in the {output_dir}/{feature_name}/tasks/ directory.\n
  • Boundary markers: The instructions do not explicitly mandate delimiters or 'ignore' instructions when passing file content to sub-agents.\n
  • Capability inventory: The skill uses the Agent tool to dispatch sub-agents capable of file modification, shell command execution (for testing), and Git operations.\n
  • Sanitization: No explicit sanitization of the file-based input is performed before interpolation into prompts.\n- [COMMAND_EXECUTION]: The workflow involves running test suites and performing Git commits. These actions are legitimate and necessary for the Test-Driven Development (TDD) process the skill is designed to facilitate.\n- [DATA_EXFILTRATION]: The skill's file operations are localized to project directories and temporary paths. Analysis found no evidence of network exfiltration, access to sensitive credentials, or unauthorized data transmission.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 12:08 AM
Security Audit — agent-trust-hub — impl