Skip to content
SPECIFICTION
Get listed

Inferwall

inferwall.com

WindowsmacOSLinuxAPI
Visit inferwall.com
Spec score 7.0 #23 of 31 in LLM Security Tools

What it is

Inferwall is a self-hosted AI firewall that scans inputs and outputs in LLM applications for threats including prompt injection, jailbreaks, data leakage, and unsafe content. It uses heuristic, machine learning, semantic, and configurable LLM judge detection layers. Deployment options are a Python SDK, standalone HTTP server, or transparent reverse proxy. Listed wrappers include OpenAI, Anthropic, LangChain, and FastAPI, with other stack integrations such as Ollama, Hugging Face, LlamaIndex, Pydantic, vLLM, and Mistral. Users can set allow, flag, or block thresholds by route and modify signatures and version policies under ~/.inferwall/. The Lite profile is about 8 MB; Standard adds about 730 MB of models, and Full adds the LLM-judge layer and requires LLM credentials. Inferwall says its Rust heuristic engine runs in under 0.3 ms at p99. The engine uses Apache 2.0; community signatures use CC BY-SA 4.0. The project cautions that detection is imperfect and positions the firewall as one part of defense in depth.

Spec sheet

Purpose
Inferwall scans LLM app inputs and outputs for prompt injection, jailbreaks, data leakage, content safety issues, and agentic threats.Read 7 Oct 2026 · inferwall.com
Deployment
It can run as a Python SDK, standalone HTTP server, or reverse proxy.Read 7 Oct 2026 · inferwall.com
Integrations
The site lists first-party wrappers for OpenAI, Anthropic, LangChain, and FastAPI.Read 7 Oct 2026 · inferwall.com
License
The engine is Apache 2.0 and community signatures are CC BY-SA 4.0.Read 7 Oct 2026 · inferwall.com
Additional ecosystem support
The site lists Ollama, Hugging Face, LlamaIndex, Pydantic, vLLM, and Mistral among supported stack integrations.Read 5 Oct 2026 · inferwall.com
Caveat
The site says no detection system is perfect and describes Inferwall as one layer in a defense-in-depth strategy.Read 7 Oct 2026 · inferwall.com
Detection layers
It combines heuristic, ML classifier, semantic, and configurable LLM-judge detection layers.Read 7 Oct 2026 · inferwall.com
Install profiles
The site describes Lite as about 8 MB, Standard as including about 730 MB of models, and Full as including those models plus LLM credentials.Read 5 Oct 2026 · inferwall.com
Licensing
The engine is licensed under Apache 2.0 and community signatures under CC BY-SA 4.0.Read 5 Oct 2026 · inferwall.com
Limitations
The site states that no detection system is perfect and presents Inferwall as one layer of a defense in depth strategy.Read 5 Oct 2026 · inferwall.com
Performance
The site says its Rust heuristic engine runs in under 0.3 ms at p99 and the Lite profile requires no GPU.Read 7 Oct 2026 · inferwall.com
Policy controls
Users can set allow, flag, or block thresholds per route and extend or override signatures under ~/.inferwall/.Read 5 Oct 2026 · inferwall.com
Profiles
The listed Lite profile is about 8 MB; Standard adds about 730 MB of models; Full adds the LLM-judge layer and requires LLM credentials.Read 7 Oct 2026 · inferwall.com
Rules
Users can override shipped signatures, add their own, and version policies under ~/.inferwall/.Read 7 Oct 2026 · inferwall.com
Security mapping
The site says its prompt injection, jailbreak, prompt extraction, and data leakage signatures map to MITRE ATLAS techniques AML.T0051, AML.T0054, AML.T0056, and AML.T0057.Read 5 Oct 2026 · inferwall.com

Strong points and limits

Strong points

  • Supports SDK, HTTP server, and reverse proxy deployment.
  • Offers configurable allow, flag, or block thresholds.
  • Lite profile is about 8 MB.
  • Engine is licensed under Apache 2.0.

Limits

  • Full profile requires LLM credentials.
  • Standard profile adds about 730 MB of models.
  • Detection is not perfect.

Who it suits

Inferwall suits teams building LLM applications that want a self-hosted layer for scanning inputs and outputs. Its deployment choices and route-level thresholds offer different ways to add that layer.

Verdict

Inferwall provides multiple deployment paths, configurable policies, and integrations for common LLM stacks. Its own guidance is clear that it should be one layer in a broader defense-in-depth approach.

Where Specifiction places it

Alternatives to InferwallAll alternatives →