The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard shell commands including rm, mkdir, git, and gh to manage project files and interact with version control systems.
[EXTERNAL_DOWNLOADS]: Interacts with Vercel (api.vercel.com) and GitHub (github.com) to facilitate the deployment and mirroring of the generated prototypes. These interactions involve well-known and trusted platforms.
[DYNAMIC_EXECUTION]: Generates application code, such as HTML pages and Node.js serverless functions, at runtime based on research data stored in the local environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local research logs and memory files to generate deployed applications.
Ingestion points: Reads from memory/MEMORY.md, memory/logs/, memory/topics/*.md, and output/articles/.
Boundary markers: No explicit delimiters or boundary markers are defined for the content read from these files to distinguish it from instructions.
Capability inventory: Access to the local file system, execution of shell commands, and the ability to perform network POST requests via curl and gh.
Sanitization: Implements a pre-flight validation step that uses grep to ensure sensitive tokens (e.g., VERCEL_TOKEN, GH_GLOBAL, sk-ant-) are not present in the code before deployment.

deploy-prototype