handoff
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill includes a dedicated section (3-A) instructing the agent to actively redact sensitive data such as API keys, tokens, credentials, and PII before writing them to disk. It suggests using placeholders like
[REDACTED]or[INTERNAL URL], which is a security best practice for managing session context. - [INDIRECT_PROMPT_INJECTION]: The skill processes session history and external artifacts to generate summaries.
- Ingestion points: Conversational history and linked project files (SKILL.md).
- Boundary markers: The output is constrained by a specific markdown template.
- Capability inventory: Contains instructions for writing files and invoking sub-agents via the
tasktool. - Sanitization: Includes specific instructions to filter and redact sensitive data from the processed context.
- [COMMAND_EXECUTION]: The instructions provide benign shell templates for managing temporary file paths, such as using
mktempin Bash or[System.Guid]::NewGuid()in PowerShell. These are used to generate random filenames for portable handoff briefs and do not involve unauthorized command execution or privilege escalation.
Audit Metadata