The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using curl, jq, and python3 to interact with the Outline API. These commands are provided as templates for performing file uploads (attachments) and as fallbacks for document updates when the primary MCP tools exhibit bugs, such as swallowing newline characters in tables.
[DYNAMIC_EXECUTION]: The skill provides Python code snippets (e.g., python3 -c "import json; ...") for the agent to generate JSON payloads and handle file I/O dynamically at runtime before sending data to the API via curl.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it ingests untrusted data from an external, potentially multi-user source (Outline Wiki) and possesses significant write and execute capabilities.
Ingestion points: Document content fetched via the fetch capability, search results from list_documents, and user comments from list_comments (documented in SKILL.md).
Boundary markers: The skill utilizes the Open Knowledge Format (OKF) and fenced code blocks for structure, but it does not specify explicit delimiters or instructions to ignore embedded commands within the fetched wiki content.
Capability inventory: The skill allows for creating, updating, moving, and deleting documents, as well as executing shell commands (curl, python3) for API interactions.
Sanitization: There are no documented procedures for sanitizing or escaping the content retrieved from the wiki before processing it in the agent's context.

yzr-outline-wiki