The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the user to modify the global ~/.claude/settings.json configuration file to permanently authorize high-privilege commands, including node, rm, sed, and browse, without manual confirmation prompts. This practice effectively bypasses the platform's security guardrails and increases the risk of unauthorized command execution if the agent is compromised.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the browse CLI utility via npm. While this is a vendor-owned resource from Browserbase, it constitutes an external dependency that increases the system's attack surface.
[PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection because its pipeline ingests and processes untrusted content from the internet.
Ingestion points: The gate_candidates.mjs script and several enrichment subagents (marketing, discussion, and news lanes) fetch content from external URLs using the browse cloud fetch tool.
Boundary markers: The instructions do not implement boundary markers or 'ignore previous instructions' warnings when passing untrusted external text into the subagent prompts.
Capability inventory: The skill utilizes the Bash tool to run custom Node.js scripts that perform file system modifications (mkdir, rm), network operations, and screenshot captures.
Sanitization: Sanitization is limited to stripping HTML tags and basic entity escaping; it does not protect against malicious natural language instructions embedded within the fetched content.

competitor-analysis