The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data from Email Service Providers (ESPs), including subscriber profiles, event history, and campaign content (subject lines and body text). This creates a surface for potential indirect prompt injection attacks if malicious instructions are embedded in the marketing data being audited.
Ingestion points: Data retrieved via ESP-specific tools such as list_flows, list_events, list_profiles, and tools for scanning campaign content.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat audited content as untrusted data or to ignore embedded instructions.
Capability inventory: The skill possesses significant capabilities, including the Bash and Write tools, as well as the ability to create findings and write to the platform context tree via Cogny MCP tools.
Sanitization: There is no evidence of sanitization or validation of the retrieved ESP content before it is processed by the agent's reasoning steps.
[COMMAND_EXECUTION]: The skill includes the Bash tool in its allowed-tools configuration, allowing for arbitrary command execution on the host environment as part of its diagnostic and environment detection workflow.

revenue-audit