startup-context
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill generates a persistent context file (
.agents/startup-context.md) based on user-provided input. This file is intended to be used by all other agent skills, creating a potential trust boundary issue where instructions embedded in startup data could influence the behavior of other skills.\n - Ingestion points: User input provided during the interview rounds defined in
SKILL.md.\n - Boundary markers: Uses structured Markdown headers but lacks specific instructions to treat content as untrusted data.\n
- Capability inventory: Writing files to the project directory.\n
- Sanitization: No sanitization of user-provided content is performed before writing to the file.\n- [DATA_EXFILTRATION]: The skill intentionally collects sensitive business and financial metrics, such as MRR/ARR, runway, and fundraising details. While this is the skill's primary purpose, users should be aware that this data is stored in plain text locally and is accessible to other agent skills.
Audit Metadata