The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the 'agent-browser' package globally from the npm registry. It also fetches instruction modules and workflows from a remote source using the agent-browser skills get command.
[COMMAND_EXECUTION]: The skill relies on executing shell commands via Bash to install the tool, manage its configuration, and perform browser automation tasks.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it is designed to ingest and process data from untrusted external sources.
Ingestion points: The skill interacts with and extracts data from arbitrary websites and Slack workspace conversations.
Boundary markers: No specific delimiters or instructions are provided to distinguish between system instructions and data retrieved from external web or Slack sources.
Capability inventory: The skill allows the execution of Bash commands, providing a significant capability that could be targeted by instructions embedded in external data.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from external sources before it is processed by the agent.

agent-browser