The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external connectors such as LinkedIn, email, and Slack to automate contact creation and profile enrichment. This creates a surface for indirect prompt injection where malicious content embedded in external messages or profiles could attempt to override the agent's instructions during the processing or logging phase.
Ingestion points: Data from third-party connectors (LinkedIn, email, Slack) and user-supplied notes about interactions are processed in SKILL.md under the 'Mode: Add new contact' and 'Mode: Log an interaction' sections.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present when processing external content.
Capability inventory: The skill utilizes file system read and write operations within the contacts/ directory.
Sanitization: There is no evidence of sanitization, filtering, or validation for the content pulled from external sources before it is interpreted by the agent.

personal-crm