The guardian angel for your codebase.

Find, validate, and fix vulnerabilities in code as it is written.

Download

From a line of code to the controls it touches.

ledgerlens/views.py

@bp.post("/report/import")def import_report():    import pickle    from flask import jsonify     report = pickle.loads(request.data)    return jsonify({"received": report})
CWE-502python-pickleCWE-345pickle-on-request-bodySOC 2CC6.8Prevention and Detection ofUnauthorized SoftwareNIST 800-53SI-3Malicious Code ProtectionSI-10Information Input ValidationSI-7Software, Firmware, andInformation Integrity
NEW / full scan
CRITICAL  ledgerlens/views.py:87
  ID: f_40f0dd7825c680fd34d51d04
  deva.cwe-345.pickle-on-request-body | CWE-345
  pickle.loads called directly on a request body / network input
  without a signature check. Either replace pickle with JSON, or
  verify a separate HMAC over the bytes before unpickling.
Verdict: Likely valid
Confidence: 99%
Explanation:
  The reachable import endpoint passes raw request bytes directly to
  pickle.loads, with no signature verification shown. Deployment
  exposure is not established.
Verdict: Fix suggested
Confidence: 94%
Proposed diff (read-only):
  @@ -82,7 +82,7 @@
   @bp.post("/report/import")
   def import_report():
  -    import pickle
  +    import json
       from flask import jsonify

  -    report = pickle.loads(request.data)
  +    report = json.loads(request.data)
       return jsonify({"received": report})
  Review redacted source lines before applying this proposed diff.
SOC2
  CC6.8 — Prevention and Detection of Unauthorized Software: covered violations
    Core finding count: 1
    Traceable report findings: 1
      Finding b630a425295d…: ledgerlens/views.py:87 | deva.cwe-502.python-pickle
        Use of pickle.load / pickle.loads on potentially untrusted data
        enables arbitrary code execution. Use json or msgpack for data
        interchange; pickle should only be used for trusted, internal
        payloads.
    Detector-covered CWEs: CWE-94, CWE-434, CWE-502, CWE-829, CWE-1321, CWE-1336
    Core evidence guidance: Show that deserialization is safe, dependencies are verified, and code injection vectors are mitigated.
Saved sarif output to ledgerlens.sarif

{
  "level": "error",
  "locations": [
    {
      "physicalLocation": {
        "artifactLocation": {
          "uri": "ledgerlens/views.py",
          "uriBaseId": "SRCROOT"
        },
        "region": {
          "endLine": 87,
          "startColumn": 14,
          "startLine": 87
        }
      }
    }
  ],
  "properties": {
    "cwe": "CWE-345",
    "deva_compliance": {
      "nist_800_53": [
        "SC-23",
        "SI-7",
        "SI-10"
      ],
      …
    },
    "severity": "CRITICAL"
  },
  "ruleId": "deva.cwe-345.pickle-on-request-body"
}

Scanning is free. Every step after deva scan needs Pro.

Preserve the context each reviewer needs.

For developers

Findings and fixes stay connected to the affected code instead of becoming detached tickets.

-  report = pickle.loads(request.data)
+  report = json.loads(request.data)

For security teams

Validation context and remediation history make review faster and more consistent.

Verdict: Likely valid
Confidence: 99%

For compliance teams

Control mappings and structured outputs create a reusable technical evidence trail.

SOC2
  CC6.8 — Prevention and Detection of Unauthorized Software: covered violations

Match the workflow to the environment.

Local analysis

Run core scanning and evidence workflows on the developer workstation.

Customer-controlled infrastructure

Keep sensitive code and analysis inside infrastructure your organization controls.

Optional AI assistance

Use supported local or cloud models when AI-assisted validation and remediation fit the environment.

Your organization

Developer workstation

$ deva scan .

CI runner

$ deva scan . --format json

Local model

$ deva models install

Cloud model, your key

$ deva providers configure openrouter
Source code stays local unless you choose a cloud model. Network details

Keep up with Deva.

New releases and product updates, by email.