asb-needs-stack
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from customer interviews and persona definition files to construct the 'Needs Stack'.
- Ingestion points: Reads content from
NEEDS-STACK.md, ideal-customer definition files, and customer-interview finding documents. - Boundary markers: The skill does not explicitly instruct the agent to use delimiters or specific ignore-instructions logic when reading external files, although it uses markdown fencing for its own generated output.
- Capability inventory: Executes file read/write operations on the local file system and invokes external search tools to identify market occupants and competitors.
- Sanitization: The skill employs a 'crystallization gate' that mandates specific, customer-centric phrasing and valid means-to-an-end links, which acts as a semantic filter against malicious instructions embedded in input data.
Audit Metadata