The Agent Skills Directory

[SAFE]: The skill is a template scaffold and does not contain malicious code. It implements safety-first patterns such as soft-failing when unauthenticated to prevent the agent from performing unauthorized actions.- [COMMAND_EXECUTION]: The scripts/check-login.mjs script executes ego-browser --help to ensure dependencies are met, which is a standard and safe operational check.- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external webpage content to detect login states. 1. Ingestion points: url, title, and text in scripts/lib/pure.mjs. 2. Boundary markers: Explicit instructions in SKILL.md define success and failure states. 3. Capability inventory: Access to ego-browser and local shell execution. 4. Sanitization: Inputs are parsed into boolean flags via regex, mitigating the risk of instructions being interpreted as commands.

__SKILL_ID__