FuzzForge
fuzzforge.ai
- Free plan
- Web
- Self-hosted
What FuzzForge is
FuzzForge is FuzzingLabs’ platform for fuzzing, emulation and reverse-engineering campaigns on embedded and connected products. It accepts binaries, complete firmware images and source code, including third-party components without source. The platform maps targets, builds an SBOM from binaries and produces triaged crashes with replayable test cases. Specialized agents run campaigns, while engineers can take over with familiar industry tools. An execution layer schedules campaigns on available compute, including runs lasting overnight or for days. Deployment options are hosted, dedicated in a customer cloud tenancy, Kubernetes on-premise or air-gapped. Air-gapped installations use local open-weight models, have no network egress and cost more than standard on-premise licensing. FuzzForge logs model requests and reasoning chains. Customers retain test cases, reports, configurations and their knowledge graph. A free plan is available.
Who it suits
FuzzForge suits product-security teams and PSIRTs, including embedded vendors, OEMs, public-sector and defense teams, MSSPs, integrators and certification bodies.
Good to know, and what to watch
- Works with binaries and firmware without source
- Produces replayable, triaged crash cases
- Builds SBOMs from binaries
- Multiple deployment models, including air-gapped
- Free plan available
- Does not replace security researchers
- Not a source-code scanner
- Air-gapped deployment costs more
- Campaigns may require extended runtimes
The verdict
FuzzForge combines automated agents with engineer control for embedded-product security campaigns. It offers broad deployment choices and retains customer campaign artifacts, while remaining a complement to researchers rather than a source-code scanner.
What the maker publishes
- Air-gapped limit
- Air-gapped deployment uses local open-weight models only, has no router or network egress, and is priced above the standard on-premise licence. source
- Air-gapped limits
- Air-gapped deployment uses local open-weight models, has no network egress, and is priced above the standard on-premise licence. source
- Audience
- The page identifies product security teams and PSIRTs as its audience, including embedded product vendors, OEMs, government and defense teams, and security service firms. source
- Audit logs
- The platform logs every model request and reasoning chain. source
- Audit logs
- The platform logs every model request and reasoning chain for auditing. source
- Automation
- Specialized agents drive campaigns, while engineers can take over using industry tools they already know. source
- Campaign results
- Campaigns produce triaged crashes with replayable test cases and an SBOM built from the binary. source
- Campaign results
- Campaigns produce triaged crashes with replayable test cases and validated findings; the platform can also build an SBOM from a binary. source
- Data ownership
- Customers keep campaign test cases, reports, configurations and their knowledge graph; FuzzingLabs keeps the platform itself. source
- Deployment
- Deployment options include a hosted instance, a dedicated instance in the customer's cloud tenancy, Kubernetes on-premise, and an air-gapped deployment. source
- Deployment
- Deployment options include hosted SaaS, a dedicated instance in the customer’s cloud tenancy, Kubernetes on-premise, and air-gapped deployment. source
- Founded
- 2021 source
- Headquarters
- Near Paris source
- Human oversight
- Agents drive campaigns, while engineers can take over using industry tools they already know. source
- Input targets
- It accepts binaries, full firmware images or source code and can map targets and build an SBOM. source
- Integrations
- FuzzingLabs describes workflows that combine tools such as Semgrep, CodeQL, AFL, LibFuzzer, Ghidra, angr and QEMU. source
- Licensing unit
- For public sector procurement, the maker describes an annual licence per instance, with tiers based on target count and no per-seat or per-test billing. source
- Maker
- FuzzingLabs describes FuzzForge as its flagship product for continuous offensive validation on firmware, binaries and embedded systems. source
- Maker
- FuzzingLabs describes FuzzForge as its flagship AI agent orchestration platform for continuous offensive validation on firmware, binaries and embedded systems. source
- No source required
- FuzzForge works on binaries and firmware images without source code and writes the fuzzing harness itself. source
- Not a source-code scanner
- The maker says FuzzForge does not replace researchers and is not a source-code scanner. source
- Operations
- The execution layer schedules campaigns on available compute, and campaigns can run overnight or for days. source
- Purpose
- FuzzForge runs fuzzing, emulation and reverse engineering campaigns for embedded and connected product security testing. source
- Scope limit
- FuzzForge does not replace security researchers and is not a source-code scanner. source
Checked against the maker's pages on 7 Oct 2026.