Open source · v0.1.0 on npm

Trace risky AI data flows before they ship.

Codelytra follows untrusted input and sensitive data through models, agents, tools, retrieval, and MCP—then shows the exact path to privileged actions.

  • Local-first
  • Deterministic findings
  • No telemetry
security scan LOCAL

$ npx codelytra scan .

Analyzing 48 files · 17 AI security rules

CRITICAL HIGH CONFIDENCE
CL-AI-001

Model output reaches command execution

01 req.body.message 02 openai.responses.create() 03 exec(response.output_text)

1 critical path found842 ms

17built-in rules
5report formats
0network calls for SAST
1command to start
Security that follows the data

Built for the paths traditional scanners miss.

Codelytra models AI-specific trust boundaries without importing, installing, building, or executing the repository it scans.

01

End-to-end flow analysis

Track user-controlled and sensitive values across functions, model boundaries, and high-impact sinks such as shells, SQL, URLs, files, and external destinations.

Evidence-rich findings

See source locations, propagation steps, confidence, severity, and focused remediation.

AI ecosystem aware

Recognizes common OpenAI, Anthropic, Vercel AI, LangChain, Google GenAI, and MCP patterns.

CI-ready outputs

Export SARIF, JSON, Markdown, HTML, and stable fingerprints for baseline-aware policies.

Optional smart explanations

Bring any OpenAI-compatible provider. AI enriches deterministic evidence—it never decides findings.

From code to action

One scan. A complete, reviewable trail.

  1. 01

    Discover

    Find AI providers, tools, inputs, sensitive sources, and privileged sinks.

  2. 02

    Trace

    Propagate trust tags through functions, prompts, retrieval, and model output.

  3. 03

    Explain

    Report the exact path, safeguards observed, and actionable remediation.

  4. 04

    Enforce

    Apply severity and confidence policies in CI with stable exit codes.

Start in seconds

Audit your AI application locally.

Node.js 22+ is the only runtime requirement. Static scans require no account, API key, network access, or telemetry.

npm install --save-dev @ezducate/codelytra
npx codelytra scan .

Optional smart features can be configured later with codelytra provider setup.

Where it fits

Useful from first review to release gate.

01

Agent development

Review tool permissions and model-controlled arguments before an agent gains production access.

02

Pull request checks

Block newly introduced high-confidence paths and upload SARIF to code scanning.

03

Security inventory

Generate an AI Security Manifest and attack graph for architecture and risk reviews.

Common questions

AI security scanning, without the black box.

What does Codelytra scan?

JavaScript and TypeScript applications that use LLMs, AI agents, RAG, tool calling, or Model Context Protocol servers.

Does a static scan send code anywhere?

No. Static analysis is local-first, makes no network requests, collects no telemetry, and never executes the target repository.

Does it detect prompt injection?

It identifies code paths where untrusted or retrieved content can influence models and privileged actions, including missing validation, authorization, or approval controls.

Can it run in GitHub Actions?

Yes. Codelytra provides stable policy exit codes, SARIF output, baselines, JSON reports, and configurable severity and confidence gates.

An EZducate open-source project

Responsible AI deserves practical security tooling.

Codelytra is maintained by EZducate, an AI-powered platform building accessible tools for educators, families, and clinicians. We are sharing the security foundations we want the broader AI ecosystem to have.

Explore EZducate

Find the path. Fix the risk.

Open source under Apache-2.0. Built for teams shipping AI responsibly.