The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from a local context file and user-provided decision details to inform its analysis. If these inputs contain adversarial instructions, they could influence the agent's behavior. The instructions lack explicit boundary markers or sanitization requirements for this untrusted data.
Ingestion points: The skill reads .agents/marketing-context.md (as specified in SKILL.md) and processes user input describing marketing decisions.
Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded commands within the ingested text.
Capability inventory: The skill utilizes web search tools (implied in references/evidence-backed-lenses.md) and file reading operations.
Sanitization: Absent; no escaping, validation, or filtering of the external content is mentioned.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to research and verify strategic frameworks online, which involves fetching data from arbitrary third-party websites.
Evidence: references/evidence-backed-lenses.md explicitly directs the agent to "Verify the sources online when current access is available."

decision-lenses