The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local Node.js script scripts/find-writing-assistance-targets.mjs to identify likely frontend integration points in the workspace. The script performs read-only file system operations, searching for specific API markers and prioritizing standard entry points like package.json and index.html.
[EXTERNAL_DOWNLOADS]: The documentation references the @types/dom-chromium-ai NPM package to provide TypeScript support for the experimental browser APIs. This is a well-known community-maintained package for Chromium AI type definitions.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of features that process untrusted data from web pages.
Ingestion points: The Summarizer, Writer, and Rewriter session methods described in SKILL.md and references/examples.md process user-provided or scraped text.
Boundary markers: The implementation templates rely on browser-native API structures which separate session-wide sharedContext from per-call input, though explicit delimiters are not enforced in the generated code.
Capability inventory: The skill enables the agent to implement browser-side AI inference capabilities (summarization, generation, and rewriting).
Sanitization: SKILL.md (Step 4.3) explicitly instructs to "Strip or normalize HTML before summarization or rewriting" to mitigate injection risks when processing rendered page content.

writing-assistance-apis