jsapi-filters
Warn
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The code snippets in SKILL.md (Step 1) and references/rpc-methods.md (Handshake) implement a window.onmessage listener that accepts a MessagePort without verifying the origin of the sender. This facilitates unauthorized cross-origin communication, as any external site can send a message to the host window to intercept the port or spoof Domo embed messages.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a data ingestion surface that processes untrusted JSON-RPC payloads from a MessagePort and performs privileged actions.
- Ingestion points: window.onmessage and port.onmessage handlers in SKILL.md that receive data from external iframes.
- Boundary markers: Absent. The logic does not verify the identity of the iframe or the integrity of the data packets.
- Capability inventory: The skill uses incoming message parameters to drive application navigation via router.push (Step 2) and perform direct DOM manipulation by setting the style.height of iframes (Step 2).
- Sanitization: None detected. Parameters such as filter values and frame dimensions are used directly in logic without escaping or validation.
Audit Metadata