Map / Runtime & Guardrails / TAWA GUARD
TAWA GUARD
tawaguard.com ↗ · LinkedIn ↗ · Open on the map
Capabilities
Pre-inference prompt-injection filter for LLM apps and AI agents: scans untrusted text (tool output, RAG chunks, retrieved documents, web pages, emails, user prompts) before the model or agent acts on it and returns allow, review or block with a risk score, per-category scores, matched spans and a rationale. Covers prompt injection, goal hijacking, jailbreaks, tool abuse and data exfiltration, plus per-source PII redact/flag/block. Dashboard shows what was flagged and why; Policy Studio for versioned rules with simulation against recorded traffic and rollback.
Sensors / integration
HTTP scan API (/v1/scan, /v1/agent/tool-output, /v1/rag/chunk) with Python and TypeScript SDKs where the app decides on the verdict, or a proxy mode that blocks directly. Detection is a cascade: heuristics on every input, deeper analysis for ambiguous ones, with de-obfuscation and language detection first; independent of any LLM. SIEM export (CEF/JSON over syslog). SaaS processed in London; self-hosting for enterprise by arrangement.
Protections
Blocks or routes to review inputs above per-surface score thresholds; redacts or blocks PII per source; audit logs and signed DPA on paid tiers. Input-side screening only, no output or agent-action enforcement described. Vendor benchmark (v13, Sep 2026): 72% attack recall at 1% false-positive budget, 23 ms median in-process latency.
Differentiator
Lightweight, LLM-independent prompt-injection gate aimed at indirect injection through content agents read, with a free developer tier (5M characters/month) and per-character pricing. Very early: Tawa Guard Ltd, London, founded 2026, ~2 employees on LinkedIn; no disclosed founders, funding or customers.
Other Runtime & Guardrails companies
AariaSec · ActaClad · ActiveFence · Acuvity · AGAT Software · AgentFox · AI Aramid · AI Prompt Shield · Aiceberg · Aim Security · AIR · Aira Security · See all →
Last updated · Get the “Listed on” badge