The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as influencer handles, niche categories, and campaign briefs, which presents a surface for indirect prompt injection attacks. Malicious instructions embedded in these inputs could attempt to influence the agent's output during pitch generation or pipeline tracking.
Ingestion points: Reads influencer handles, platform data, and campaign context from user inputs and existing memory files located at memory/creators/.
Boundary markers: The skill explicitly instructs the agent to label all outreach artifacts as DRAFT — NOT SENT and requires the user to explicitly approve recipients and messages before any side effects occur.
Capability inventory: The skill can write files to the local memory/ directory and update registry logs through calls to the registry-events.py script.
Sanitization: The instructions do not specify any validation or sanitization techniques to prevent embedded instructions in external data from being interpreted as commands.

outreach-manager