The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core function is to ingest untrusted data from various social media platforms (X, LinkedIn, Reddit, etc.) and convert it into structured JSON. This external content is then processed by the agent, creating a surface for indirect prompt injection.
Ingestion points: Data is retrieved from external URLs using curl, WebFetch, or browser subagents as defined in SKILL.md and references/strategies.md.
Boundary markers: The skill does not provide instructions for the agent to use specific delimiters or to disregard instructions contained within the fetched content.
Capability inventory: The agent has access to curl for network requests and file system access to write data to ~/Documents/social-fetches/.
Sanitization: While the skill normalizes data into a schema, it lacks specific instructions for sanitizing or escaping potentially malicious text before it is returned to the agent context.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill facilitates network communication with several non-whitelisted third-party domains and manages sensitive authentication tokens.
Network operations: The skill instructs the agent to send requests to api.fxtwitter.com, nitter.net, archive.org, api.scrapecreators.com, and api.apify.com.
Credential handling: The skill correctly advises using environment variables ($SCRAPECREATORS_API_KEY, $APIFY_API_TOKEN) rather than hardcoding keys, but it does instruct the agent to include these secrets in network request headers to third-party services.
Local file access: The --save and --media flags instruct the agent to write data to the user's local directory (~/Documents/social-fetches/).
[COMMAND_EXECUTION]: The skill provides numerous shell command templates using curl, jq, and grep for the agent to execute at runtime.
Evidence: references/strategies.md contains multiple pre-defined shell commands for interacting with platform APIs and mirrors.

social-fetch