The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages sensitive browser session state files located in ~/.notis-agent-browser/. These files contain cookies and authentication tokens which are treated as credential material by the skill.
[CREDENTIALS_UNSAFE]: The skill uses the op (1Password) CLI to retrieve sensitive usernames and passwords using service account tokens.
[REMOTE_CODE_EXECUTION]: The skill uses the agent-browser skills get command to dynamically download and load instructional content from a remote server at runtime, which influences agent behavior.
[EXTERNAL_DOWNLOADS]: The skill installs the agent-browser package globally from npm using npm i -g agent-browser.
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform all browser automation tasks, state management, and secret retrieval.
[INDIRECT_PROMPT_INJECTION]: The skill has a large attack surface for indirect prompt injection as it processes live website content.
Ingestion points: Website content is ingested via agent-browser snapshot -i in SKILL.md.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the provided snapshots.
Capability inventory: The skill possesses high-privilege capabilities including shell execution, file system writes to sensitive paths, and secret manager access.
Sanitization: No sanitization of the accessibility tree or page content is performed before processing.

notis-browser-control