The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the notte CLI to manage the lifecycle of functions, including listing, creating, updating, and deleting functions. It also uses the CLI to manage cron-based scheduling and session-based browser automation.
[EXTERNAL_DOWNLOADS]: Function source code is downloaded via curl using URLs dynamically retrieved from the Notte platform's API (notte functions show). This is a standard procedure for retrieving function assets for local inspection and patching.
[DATA_EXPOSURE]: The skill provides instructions for accessing environment secrets via notte functions secrets and run metadata via notte functions run-metadata. While this involves sensitive data, the operations are restricted to the authenticated user's environment and are essential for the primary purpose of diagnosing functional failures.
[INDIRECT_PROMPT_INJECTION]: Because the skill involves reading and re-exploring web content that may have changed, it is potentially subject to indirect prompt injection if the external data contains malicious instructions. The skill explicitly mitigates this by instructing the agent to treat page content as untrusted and requiring a confirmation gate before promoting any code changes.

notte-functions-doctor