The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documentation indicates it is designed to ingest and process untrusted data from X/Twitter, such as brand mentions, search results, and follower biographies. This creates an attack surface where instructions embedded in social media content could influence the agent's behavior.
Ingestion points: Profile pages, follower lists, and search results on x.com (referenced in SKILL.md).
Boundary markers: No delimited zones or "ignore instructions" warnings are documented for processing external text.
Capability inventory: Documentation describes browser automation (Puppeteer) for brand monitoring, audience analysis, and automated replies.
Sanitization: There is no mention of content sanitization, filtering, or validation for external data.
[NO_CODE]: The documentation references multiple functional scripts in a src/ directory (including businessTools.js, tweetABTester.js, autoPlugReplies.js, audienceDemographics.js, and tweetPerformance.js), but the actual source code for these files is missing from the skill package.

business-ads