The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to search, read, and synthesize information from external knowledge base repositories (Markdown, Text, HTML). These files represent untrusted data that could contain malicious instructions designed to influence the agent's behavior during synthesis.
Ingestion points: Reads content from files located in paths defined in ~/.superskills/knowledge.conf (SKILL.md).
Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore instructions embedded within the knowledge base files.
Capability inventory: The skill has access to Bash, Write, Read, Grep, and Glob tools. It can execute git commands and write new files (SKILL.md).
Sanitization: Absent. The skill quotes passages and connects principles to user situations without explicit sanitization or escaping of the input content.
[COMMAND_EXECUTION]: The skill uses shell commands via the Bash tool to interact with the file system and version control.
Evidence: Uses cat, find, grep, awk, and git (add, commit, pull) to manage knowledge bases. While these are restricted to the configured paths, they provide a capability that could be misused if the configuration file is tampered with.

kb-advisor