Code analysis

static source read inferred

Static code-analysis findings — hidden prompt content in shipped skill files, committed secrets, dynamic-exec sinks, and suspicious call-home endpoints — across the analyzed catalogue. Heuristic, pure, no code executed; every row deep-links to its source. Click a kind to filter.

analysis coverage 85% of 59,140 analyzable servers
50020 analyzed
8131 re-analysis due
988 not analyzable
1 not yet analyzed
4803 source gone
not analyzable
789 too large 199 no source

Running analyzer v33. The scanner changelog explains what each version detects and when it changed.

code findings 15 shown
  1. HIGH dynamic exec PrefectHQ/fastmcp eval()/exec() exec(compile(example.source, str(example.path), "exec"), namespace)
  2. HIGH dynamic exec PrefectHQ/fastmcp eval()/exec() exec(compile(matches[0].source, str(path), "exec"), namespace)
  3. HIGH dynamic exec PrefectHQ/fastmcp __import__() __import__(alias.name)
  4. HIGH dynamic exec PrefectHQ/fastmcp unsafe yaml.load() parsed = yaml.load(frontmatter_text, Loader=yaml.BaseLoader)
  5. HIGH committed secret cpaumelle/docplane AWS access key id AKIAZZ…(20 chars, redacted)
  6. HIGH committed secret cpaumelle/docplane AWS access key id AKIAIO…(20 chars, redacted)
  7. HIGH dynamic exec cpaumelle/docplane __import__ sink @pytest.mark.skipif(not __import__("os").environ.get("DB_HOST"), reason="requires a PostgreSQL database")
  8. HIGH committed secret cpaumelle/docplane AWS access key id AKIAIO…(20 chars, redacted)
  9. MEDIUM suspicious endpoint harness/mcp-server 2.3.4.5 { cluster: "prod", url: "https://2.3.4.5" },
  10. MEDIUM suspicious endpoint harness/mcp-server 1.2.3.4 { cluster: "staging", url: "https://1.2.3.4" },
  11. MEDIUM suspicious endpoint harness/mcp-server 1.2.3.4 " {list:{elements:[{cluster:'staging', url:'https://1.2.3.4'}, {cluster:'prod', url:'https://2.3.4.5'}]}}\n" +
  12. MEDIUM suspicious endpoint harness/mcp-server docs.datadoghq.com (telemetry) te/region identifier. Valid values (per https://docs.datadoghq.com/getting_started/site/): \"datadoghq.com\" (US1), \"us3.datadoghq.com\" (US3), \"us5.datadoghq.com\" (US5), \"datadoghq.eu\" (EU1), \"
  13. MEDIUM suspicious endpoint harness/mcp-server docs.datadoghq.com (telemetry) datadoghq.eu", "ap1.datadoghq.com"; see https://docs.datadoghq.com/getting_started/site/), datadogCredentialsSecretName (REQUIRED on Kubernetes — name of k8s secret containing "dd-api-key" and "dd-app
  14. HIGH hidden prompt agentskillexchange/skills skill: skill-exfil secret→sink: - **API key:** set `ZINC_API_KEY` from [app.zinc.com](https://app.zinc.com) and send `Authorization: Bearer $ZINC_API_KEY` to `/orders` and related endpoints. This is the standard flow fo
  15. HIGH hidden prompt agentskillexchange/skills skill: skill-exfil secret→sink: - npx add-mcp https://mcp.neon.tech/mcp --header "Authorization: Bearer <$NEON_API_KEY>"