The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to its core function of ingesting untrusted external data.
Ingestion points: Untrusted content enters the agent's context from user-provided URLs processed by web_fetch, crwl, and browser-use as described in SKILL.md and references/tool-catalog.md.
Boundary markers: Absent; the instructions do not define delimiters or provide specific directives for the agent to ignore instructions embedded within the fetched content.
Capability inventory: The skill possesses network access via crawling tools and returns the extracted content for further processing or display.
Sanitization: Absent; the skill extracts Markdown content from the source without sanitizing for malicious prompts or hidden instructions.
[COMMAND_EXECUTION]: The skill invokes external CLI tools (crwl and browser-use) using user-provided URL arguments to perform its extraction logic.
[COMMAND_EXECUTION]: The skill uses dynamic JavaScript evaluation (browser-use eval) to interact with the DOM and extract specific text elements such as titles and article bodies.

fetch