The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous Python and shell command examples (using urllib.request and curl) intended for the agent to use to interact with external APIs through the gateway.
[DATA_EXFILTRATION]: The skill is designed to transmit user data to over 100 external service providers and proxies all traffic through the author's domains (maton.ai, gateway.maton.ai).
[PROMPT_INJECTION]: The skill facilitates a broad surface for indirect prompt injection attacks.
Ingestion points: The agent is instructed to read data from 100+ external sources (e.g., references/slack/README.md, references/google-mail/README.md, references/notion/README.md), which may contain attacker-controlled instructions.
Boundary markers: Absent. The instructions do not define delimiters or provide warnings to ignore embedded commands in the ingested data.
Capability inventory: The skill possesses extensive capabilities including network operations and CRUD access to diverse platforms (e.g., references/github/README.md, references/stripe/README.md).
Sanitization: Absent. No logic is provided to sanitize or filter the data received from external APIs before it is processed by the agent.

api-gateway