The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python orchestrator script to manage research tasks. It mitigates command injection risks by writing the user-provided topic to a temporary file before reading it into the script, rather than passing it directly through a shell command line.\n- [EXTERNAL_DOWNLOADS]: The skill performs network requests to fetch data from Reddit and X. These operations target well-known, established services and are necessary for the skill's primary research function.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from social media and web pages, which presents an attack surface for indirect prompt injection.\n
Ingestion points: Data enters the system via scripts/lib/openai_reddit.py, scripts/lib/xai_x.py, and the agent's internal web search tool.\n
Boundary markers: The skill's instructions in SKILL.md explicitly direct the agent to treat all fetched content as untrusted evidence and to disregard any instructions found within the data.\n
Capability inventory: The skill can perform network operations and write reports and cache files to the local file system.\n
Sanitization: Content is parsed using JSON and regex before being presented to the agent, providing a layer of normalization that isolates the raw data.

last30days