The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external social platforms and news sources.
Ingestion points: External mentions are fetched via hn.py, bluesky.py, fediverse.py, discourse.py, gdelt.py, and tavily.py. It also processes user-provided social media exports.
Boundary markers: The skill contains explicit instructions to treat all fetched content as untrusted input per the project's security guidelines.
Capability inventory: The skill writes reports to memory/social/ and submits mention events to memory/events/channels.ndjson via a local script (registry-events.py).
Sanitization: There are explicit constraints stating that text inside a mention can never reclassify itself, suppress crisis flags, or modify the query set.
[EXTERNAL_DOWNLOADS]: The skill connects to various external search and social media services to perform its primary function.
Fetches data from Algolia (HN Search), Bluesky, GDELT (global news), and Tavily (web search).
These operations are scoped to the skill's purpose of brand monitoring.
[COMMAND_EXECUTION]: The skill interacts with local python scripts to perform network operations and record events.
Executes connector scripts in scripts/connectors/ to interface with external APIs.
Invokes registry-events.py using an operation: propose request to update the channel registry.

social-pulse-monitor