The Agent Skills Directory

[SAFE]: The skill performs text analysis and generation based on local file inputs (social-context.md and platform-rules.md). It does not demonstrate any patterns associated with data exfiltration, command execution, or malicious obfuscation.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface by ingesting data from social-context.md (located at the project root or .agents/social-context.md) to guide the agent's output. While this file influences the prompt, the skill lacks high-risk capabilities such as network access, file writing, or shell execution, limiting the potential impact of any injected instructions.
[INGESTION_POINTS]: Untrusted data may enter via social-context.md or .agents/social-context.md.
[BOUNDARY_MARKERS]: The skill does not define explicit boundary markers or 'ignore' instructions for the content of the context files.
[CAPABILITY_INVENTORY]: The skill is limited to reading files and generating text responses; it contains no subprocess calls, network operations, or file-write capabilities.
[SANITIZATION]: No explicit sanitization or filtering of the ingested context data is performed.

social-post