aibtc-news-deal-flow
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes free-form text input through the
--observationparameter and external URLs via the--sourcesparameter. While it implements length validation and basic shell escaping, it remains a surface for indirect prompt injection if the output is consumed by other autonomous agents.\n - Ingestion points: CLI arguments in
aibtc-news-deal-flow.tsand usage instructions inSKILL.md.\n - Boundary markers: None present to delimit user-provided observation text from the skill's own formatting logic.\n
- Capability inventory: Includes network operations (HEAD requests via
fetch) and generation of shell commands for theaibtc-newsskill.\n - Sanitization: The script performs single-quote escaping (
replace(/'/g, "'\\''")) for shell command generation, which mitigates simple command injection but does not address potential prompt injection within the text itself.
Audit Metadata