The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs shell commands (e.g., grep, ls, find) by interpolating user-provided feature names directly into the command strings. This creates a surface for shell command injection if the input is not sanitized by the platform.
[COMMAND_EXECUTION]: The skill utilizes dynamic execution of Python snippets (python3 -c) via the Bash tool to parse and manipulate local JSON configuration schemas during the research phase.
[DATA_EXFILTRATION]: The skill is configured to read sensitive codebase files, including Go backend handlers and configuration schemas, while also having access to outbound network tools like WebFetch and WebSearch. This combination creates a technical capability for data exfiltration if the agent is misdirected.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core workflow of ingesting and processing untrusted data.
Ingestion points: Data enters the agent context from external websites via WebFetch, library documentation via mcp__context7__query-docs, and existing local MDX documentation.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious instructions embedded within the researched materials.
Capability inventory: The skill has broad capabilities including file system modification (Write, Edit), shell execution (Bash), and task management.
Sanitization: There is no specified logic to sanitize or validate external content before it is used to generate or review documentation.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch external documentation and search the web using WebFetch, WebSearch, and specialized MCP tools for library resolution.

docs-writer