The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (blog posts, transcripts, outlines) to generate content drafts.
Ingestion points: External source content is read via the --source argument in scripts/generate_batch.mjs or through direct chat input.
Boundary markers: The instructions in SKILL.md and content/busa-writer-app/_node.json include explicit safety constraints, instructing the agent to never invent statistics, claims, or endorsements not present in the source material.
Capability inventory: The skill has the capability to write to the Busabase API and the local filesystem (via the export script).
Sanitization: The companion web application uses escapeHtml functions in app.js and writer-model.js to mitigate cross-site scripting when rendering draft content.
[COMMAND_EXECUTION]: The skill requires the execution of local Node.js scripts for its primary workflow.
The agent is instructed to run node scripts/generate_batch.mjs to create drafts and node scripts/export_decisions.mjs to package approved items.
scripts/publish_airapp.mjs and scripts/setup.mjs are used for deployment and workspace initialization.
A human-in-the-loop requirement is enforced, as the export_decisions.mjs script only processes items marked as approved by a user in the review UI.
[EXTERNAL_DOWNLOADS]: The project relies on external Node.js dependencies.
Dependencies include busabase-sdk, hono, and @hono/node-server.
These resources are managed via standard package managers (pnpm) and originate from the skill author (Busabase) or well-known, trusted open-source maintainers.

busa-writer