unstuck
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input regarding roadblocks and goals, which is then interpolated into its internal workflow and written to a local archive.
- Ingestion points: Step 1 (Goal, Wall, Attempts) in SKILL.md.
- Boundary markers: The skill follows a rigid 7-step logical workflow which provides inherent structure, though no specific character-based delimiters are enforced for input sanitization.
- Capability inventory: File system write access to the user-specific ~/.config/makerskills/ directory.
- Sanitization: Input data is treated as descriptive text and logged without explicit escaping or filtering.
- [SAFE]: The archival functionality targets a standard, non-sensitive configuration path (~/.config/makerskills/unstuck/archive/) that does not overlap with critical system files, credentials, or SSH keys. The core logic consists of conceptual problem-solving techniques and does not involve remote code execution, package installation from unknown sources, or network operations.
Audit Metadata