The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple external sources, creating an attack surface where malicious instructions embedded in social content could influence the agent's behavior.
Ingestion points: SKILL.md describes gathering data from X (Twitter) mentions, Discord messages, YouTube comments, and Brave Search results to inform the drafting of social media "hooks."
Boundary markers: The instructions do not specify the use of delimiters or clear "ignore instructions" prompts when processing external content.
Capability inventory: SKILL.md utilizes tools to write to local context nodes, create review tickets, and schedule posts on social media platforms (LinkedIn and X).
Sanitization: There is no evidence of sanitization or filtering of the ingested external text before it is used for prompt generation.
[EXTERNAL_DOWNLOADS]: The skill fetches data from several well-known external platforms and CRM services to perform its functions.
Evidence: Accesses APIs for LinkedIn, X, YouTube, Discord, Brave Search, HubSpot, Pipedrive, and Crisp.
Analysis: These interactions are conducted through the Cogny MCP gateway, which is the official infrastructure provided by the skill's author.

social-engine