sales-woocommerce
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements a feedback loop by instructing the agent to append newly discovered insights ("gotchas") to
references/learnings.md. This file is read at the start of every session to provide context. This mechanism creates a surface for indirect prompt injection where malicious or misleading instructions could be persisted across sessions if the agent is successfully manipulated during a conversation. - Ingestion points: User-provided prompts and the
references/learnings.mdfile (read in Step 1). - Boundary markers: No delimiters are specified to separate user-originated "learnings" from the rest of the authoritative reference data.
- Capability inventory: The skill provides instructions for reading and appending to local markdown files.
- Sanitization: There are no instructions to sanitize or validate the content before it is appended to the learnings file.
- [SAFE]: The skill uses placeholders (e.g.,
ck_live_xxx,cs_live_yyy) for API credentials in all code examples and documentation, adhering to security best practices. - [SAFE]: All external references point to well-known WooCommerce domains or their official GitHub organization, which are established and reputable service providers.
Audit Metadata