The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use grep to identify backlinks between articles. This is a standard search operation performed on local text files within the wiki directory and does not pose a security risk in this context.
[PROMPT_INJECTION]: The skill exhibits an ingestion surface for indirect prompt injection by processing external markdown files.
Ingestion points: The agent reads content from raw/, wiki/, and local/ directories.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to ignore embedded commands within the ingested files.
Capability inventory: The agent can read files and execute local grep commands for search.
Sanitization: Absent; the content is processed as plain text without validation or filtering of potential injection patterns.

ml-llm-wiki