The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with the local environment using tools like uxc, x-webmcp-cli, and npx. It provides a script ensure-links.sh to configure links to the underlying web bridge.
[EXTERNAL_DOWNLOADS]: The skill documentation includes instructions to download browser binaries via npx playwright install, which is a standard procedure for browser automation tools.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks because it reads content from X and possesses the ability to post data back to the platform.
Ingestion points: External data is ingested through tools like timeline.home.list, search.tweets.list, and grok.chat as specified in SKILL.md.
Boundary markers: No boundary markers or special instructions are defined to protect against instructions embedded in fetched X content.
Capability inventory: The skill includes tools for creating and replying to tweets (tweet.create, tweet.reply) and publishing markdown articles.
Sanitization: There are no explicit sanitization steps mentioned for the data retrieved from X APIs or web pages.

x-webmcp