The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources and incorporates it into the application's source code.
Ingestion points: External tweet content is fetched from api.fxtwitter.com (via scripts/fetch-tweet.mjs) and publish.twitter.com (via curl).
Boundary markers: The instructions lack explicit delimiters or warnings to treat the fetched tweet content as data only, increasing the risk that embedded instructions in a tweet could be followed by the agent.
Capability inventory: The skill has the capability to write to local source files (apps/web/lib/testimonials.ts) and execute shell commands (node, curl, npx).
Sanitization: While the script performs basic whitespace trimming and prefix removal, there is no robust sanitization or validation of the tweet content before it is proposed for insertion into the codebase.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to fetch data and verify the build.
Evidence: Execution of node .agents/skills/add-x-tweet/scripts/fetch-tweet.mjs, curl for oEmbed data, and npx tsc for TypeScript verification.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch data from third-party services.
Evidence: The scripts/fetch-tweet.mjs script makes network requests to api.fxtwitter.com. The fallback workflow uses curl to fetch data from publish.twitter.com (a well-known service).

add-x-tweet