openclaw-behavior-validator
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with external applications, APIs, and CLI tools which serves as a potential vector for indirect prompt injection.
- Ingestion points: The skill parses user-provided 'behavior contracts' (defined in
references/contract-template.md) and interacts with external targets specified within those contracts. - Boundary markers: The skill instructions recommend using an isolated temporary workspace directory to prevent data contamination but do not specify explicit delimiters for the contract text itself.
- Capability inventory: The skill is capable of executing CLI commands, performing network requests (API/Web), and managing a local temporary file system.
- Sanitization: The skill contains explicit instructions to redact credentials, tokens, cookies, and private user data from all logs, evidence, and reports.
- [COMMAND_EXECUTION]: The
SKILL.mdinstructions include shell commands for environment setup, such as creating a restricted temporary directory usingmktempandchmod 700, and copying files. These are standard automation tasks for establishing a secure, isolated workspace.
Audit Metadata