amaran-lights

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). In src/amaran_cli/mesh.py, runtime BLE advertisements and GATT notifications (from nearby devices on the user’s Bluetooth mesh proxy service) are parsed/decrypted and can reach the agent’s logic via decoded status values.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged literal high-entropy cryptographic keys found in the test vectors. tests/test_vectors.py defines NET_KEY and APP_KEY as 32-hex-digit strings (128-bit keys) which are directly usable by MeshController to derive encryption/privacy material; the file also contains the expected derived enc/priv hex values. These are not placeholders or simple example passwords — they are real-looking mesh/app keys (high entropy) and therefore qualify as secrets. No other files contain comparable high-entropy credentials; environment variable names and simple strings elsewhere are documentation or non-secret.