The Agent Skills Directory

[SAFE]: The skill is functionally focused on legal contract review and uses standard tools to retrieve internal context from Slack, Notion, and Gmail. All operations are consistent with the stated purpose of assisting with legal document analysis for startups.- [PROMPT_INJECTION]: The skill presents an inherent surface for indirect prompt injection as it processes external contract documents provided by users. An attacker could potentially include malicious instructions within a contract text to manipulate the resulting risk assessment. This is a common risk for AI agents processing untrusted text and is addressed by platform-level guardrails. 1. Ingestion points: Ingests external contract documents (PDF/text) as described in Step 1. 2. Boundary markers: The skill does not define specific delimiters for separating untrusted document content from instructions. 3. Capability inventory: Accesses internal company data via search tools (search_all_summaries, search_notion, search_emails). 4. Sanitization: No explicit sanitization or instruction-filtering logic is present for the input documents.

contract-review