The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides Python and Shell snippets for interacting with the Bright Data API. These involve standard network operations using curl and urllib.request to api.brightdata.com. These are legitimate for the skill's purpose and include explicit instructions to the agent to avoid paid endpoints (/webarchive/dump) without user consent.
[CREDENTIALS_SAFE]: The skill correctly instructs the agent to never hardcode API tokens, instead recommending environment variables or local gitignored files. While it provides code examples that read from specific local paths (e.g., ~/projects/batm/scratch/brd_token.txt), these are presented as fallbacks for the user's environment rather than credentials provided by the skill itself.
[SAFE]: The skill implements a 'cost enforcement' policy, requiring the agent to perform free searches first and display the calculated cost to the user before initiating any paid 'dump' operations. This is a safety best practice for skills interacting with paid APIs.

bright-data