The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from X/Twitter conversations and profiles using tools like x_search_tweets and x_get_tweets which could contain hidden or malicious instructions.
Ingestion points: Workflow steps in SKILL.md process user tweets and bios.
Boundary markers: No boundary markers or instructions to ignore embedded commands are present to protect the agent.
Capability inventory: The skill description suggests using automated engagement and following actions through scripts like src/engagementBooster.js and src/automation/keywordFollow.js.
Sanitization: No sanitization or data validation logic is provided for the external content.
[NO_CODE]: The skill references multiple scripts (src/keywordMonitor.js, src/audienceDemographics.js, src/audienceOverlap.js, src/engagementBooster.js, src/automation/keywordFollow.js) that were not included in the provided files. Because these scripts perform automated browser actions, their safety and actual functionality cannot be verified.

lead-generation