The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation instructs the installation of prerequisite tools from a non-trusted GitHub repository (https://github.com/23blocks-OS/ai-maestro-plugins.git).
[REMOTE_CODE_EXECUTION]: The installation process involves cloning a remote repository and immediately executing a shell script (./install-doc-tools.sh), which is a high-risk pattern when the source organization is not a recognized trusted vendor.
[COMMAND_EXECUTION]: The skill defines several commands (docs-search.sh, docs-index.sh, docs-get.sh, etc.) that execute shell scripts to perform documentation operations.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection because its core behavior involves reading and interpreting documentation content (JSDoc, docstrings, READMEs) which may be controlled by third parties. The skill explicitly directs the agent to follow a 'Search docs -> Then implement' flow, which could lead the agent to execute malicious instructions embedded in codebase comments.
Ingestion points: Documentation data retrieved via docs-search.sh and docs-get.sh in SKILL.md.
Boundary markers: None present; the agent is not instructed to ignore or treat documentation content as untrusted data.
Capability inventory: The skill allows the agent to execute multiple search, retrieval, and indexing scripts across the file system.
Sanitization: None detected in the documentation; documentation strings are processed directly as context for implementation.

docs-search