auto-improve-tests

Pass

Audited by Gen Agent Trust Hub on May 3, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from user source files and existing test code. This ingestion provides a pathway for malicious instructions to manipulate the agent's behavior during the iterative test generation process.
  • Ingestion points: Target source files and existing test files as specified by the user (SKILL.md).
  • Boundary markers: Absent; the agent lacks specific instructions or delimiters to disregard embedded commands or override attempts within the provided source code.
  • Capability inventory: The agent possesses the capability to write files (test files) and execute shell commands (test runners like Vitest or Jest) as part of its core loop.
  • Sanitization: Absent; no input validation, escaping, or structural analysis is applied to the input source code before it is used to generate or review tests.
  • [COMMAND_EXECUTION]: The skill's primary workflow requires the agent to execute test runners on newly generated code. This creates a potential security risk where a compromise via prompt injection in the source files could lead to the execution of arbitrary malicious payloads during the test phase.
Audit Metadata
Risk Level
SAFE
Analyzed
May 3, 2026, 06:06 AM
Security Audit — agent-trust-hub — auto-improve-tests