Run it yourself
open·kritt
FreeAGPL-3.0
Self-host the orchestration and choose your models, workflows, repositories, and severity rules.
- Local or remote repositories
- Bring Codex or Claude Code
- Custom workflows and agent skills
Open source orchestration · managed research
Find vulnerabilities thatsurvive review.
Kritt breaks a codebase into focused security tasks, runs AI agents in parallel, and turns their work into ranked, verified findings. Self-host the open-source engine or let our researchers run it for you.
License
AGPL-3.0
Targets
Local + GitHub
Output
Ranked findings
release-candidate-42
security workflow
0112 tasks
Map attack surface
02running
Fan out agent research
033 candidates
Verify exploit paths
041 high
Rank + de-duplicate
highconfidence 0.92
State transition can bypass finality check
Verified path · duplicate check passed · report draft ready
Open source · AGPL-3.0
The research loop,made inspectable.
open·kritt is the open-source distillation of our internal workflow: decompose the codebase, run agents against focused tasks, and turn their output into ranked findings you can inspect and extend.
$ git clone https://github.com/Kritt-ai/open-kritt
$ cd open-kritt
$ cd open-kritt
$ ./kritt setup
✓ model access configured
✓ local workspace ready
✓ model access configured
✓ local workspace ready
$ ./kritt start
→ UI running at localhost:5173
→ UI running at localhost:5173
Focused workflows
Turn a broad review into small, typed prompt steps that agents can reason about well.
Parallel agent runs
Fan work out across functions, files, and hypotheses, then recombine the useful results.
Verification passes
Run post-scripts per finding to validate exploitability, build a PoC, or draft a report.
Your severity rules
Rank, normalize, and de-duplicate findings against the impact criteria that matter to you.
Choose your path
Use the engine your way.
Run open·kritt on your own infrastructure, or have our security research team operate a managed scan for you. Managed engagements start with a short call to align on scope, authorization, and timing.
Kritt managed
Single scan
$5,000up to 200k LOC
We run the research workflow and deliver a written report of the high-impact findings we validate.
- Written findings report
- Email support during the scan
- Scope and authorization review
Teams & continuous coverage
Enterprise
Contact us
Tailored security research for teams that need CI/CD, release coverage, custom volumes, and SLAs.
- Dedicated support
- PR and release scanning
- Custom volumes and SLAs
Projects & ecosystems
Helped secure production code across leading networks.
Research track record
Built from real security work.
Kritt was shaped by the same workflows our team uses in public bug-bounty research. The Blockian record shows the kind of verified, high-impact work the system was built to support.
Managed research
Start with a conversation, not an account.
Book a demo and we will scope the engagement directly with you before any work or payment begins.
01
Book a demo
Choose a time and tell us what kind of code security problem you are trying to solve.
02
Share the scope
We align on repositories, languages, lines of code, authorization, priorities, and timing.
03
Confirm the engagement
You receive a clear plan covering deliverables, commercial terms, and the path to kickoff.
One next step
Book a demoBook a demo
Talk directly with the team that will scope and operate the research, from initial priorities through kickoff and reporting.
FAQ
Straight answers.
What changes between running the open-source platform yourself and asking Kritt to run a managed scan.
Your next security run
Bring us your hardest codebase.Let’s scope the right engagement.
Tell us what you are protecting, where the risk is concentrated, and what success looks like. We will map the right managed research plan with you.