cerbos-synapse-extension
Fail
Audited by Snyk on Aug 25, 2026
Risk Level: CRITICAL
Full Analysis
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 1.00). The skill tells the agent to collect/substitute the distribution repo and to run
docker login ... --password=YOUR_LICENCE_KEY, which requires embedding a user’s license key/password verbatim into generated commands (an insecure secret-handling pattern).
CRITICAL E005: Suspicious download URL detected in skill instructions.
- Suspicious download URL detected (high risk: 0.70). Yes — several URLs point to arbitrary cloud storage or script-hosting endpoints (S3, GCS, and a custom scripts host) that can deliver executable artifacts (.wasm or code) from untrusted sources and are therefore higher-risk distribution vectors.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). Synapse’s routeExtensions (Starlark) ingest arbitrary outsider HTTP request fields (e.g.,
req.path,req.query_params,req.body) at runtime viahandleHTTPRoute(req)under/ext/..., which are then processed by the extension/LLM logic.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). Multiple files show that Synapse can load extensions over networked URLs at runtime (e.g. "wasm+s3::https://s3.amazonaws.com/bucket/ext.wasm", "starlark+gcs::https://www.googleapis.com/storage/v1/bucket/ext.star", "wasm+git::ssh://git@github.com/org/repo//ext.wasm?ref=v1.0.0" and "starlark+https://scripts.example.com/ext?checksum=sha256:..."), and the build pipeline Dockerfile fetches and extracts remote release assets from GitHub (https://api.github.com/.../releases/latest), so remote content is fetched during build/startup and can supply/execute extension code.
Issues (4)
W007
HIGHInsecure credential handling detected in skill instructions.
E005
CRITICALSuspicious download URL detected in skill instructions.
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W012
MEDIUMUnverifiable external dependency detected (runtime URL that controls agent).
Audit Metadata