project-knowledge-harness
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill implements its functionality through a series of Bash scripts that manage file-based project memory. These scripts perform routine operations such as file creation, appending text snippets, and basic text processing using standard utilities like awk and sed.
- [INDIRECT_PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection by processing data from project files. 1. Ingestion points: sweep-inbox.sh reads backlog/inbox.md. 2. Boundary markers: Absent in the input file; the script relies on line-by-line parsing of markdown list items. 3. Capability inventory: File writes (TODO.md, backlog/, pitfalls/), local command execution (td CLI). 4. Sanitization: Title strings are validated for formatting, and shell variables are quoted to prevent command injection.
- [SAFE]: No malicious patterns, such as data exfiltration, obfuscation, or unauthorized privilege escalation, were detected. The skill is a legitimate productivity tool for software development workflows.
Audit Metadata