The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of retrieving and synthesizing untrusted data from various internet sources.
Ingestion points: The skill ingests untrusted data from Reddit threads, X posts, and general web articles via web_search, reddit_search, bird search, and web_fetch (SKILL.md, SKILL-OC.md).
Boundary markers: The output synthesis templates do not utilize specific delimiters or instructions to prevent the agent from executing commands that might be embedded in the retrieved content.
Capability inventory: The skill incorporates network-access capabilities for searching and fetching article data from the web (SKILL.md).
Sanitization: There is no mention of sanitization, filtering, or validation of external content before it is incorporated into the research reports.
[COMMAND_EXECUTION]: The skill executes shell commands using the bird CLI tool to perform data retrieval from X/Twitter.
[EXTERNAL_DOWNLOADS]: The skill depends on external resources including the Brave Search API and the Bird CLI for its primary functionality.
[DATA_EXFILTRATION]: The skill performs network operations to well-known services such as Reddit, X, and the Brave Search API. It also references the local path ~/.openclaw/credentials/bird.env for environment configuration, which is a standard practice for tool authentication in this environment.

last30days