io.github.ECD5A/tkach-securityai

Tkach Security

Verified · today

Fail-closed MCP adapter for untrusted model output over a local Tkach runtime.

Our take

Wraps untrusted model output in a fail-closed adapter backed by a local Tkach runtime, aiming to reduce unsafe execution paths. It is for developers building security-conscious agent pipelines, though the small footprint means buyers should validate the runtime and threat model carefully before relying on it.

reviewed by hand · 2026-09-13

Something wrong or dead here? Report it

Verification record

last verified
today
github stars
1
last commit
today
archived
no
license
Apache-2.0
in registry since
2026-09-12

github.com/ECD5A/Tkach-Security

More in ai

browse all →
claude-flowai

claude-flow

Verified · 26 days ago

AI orchestration with hive-mind swarms, neural networks, and 87 MCP tools for enterprise dev.

hand-reviewed68k starschecked 26 days ago

tldrawai

tldraw

Verified · 28 days ago

Draw and visually collaborate with your agents on tldraw's canvas.

hand-reviewed50k starschecked 28 days ago

codebase-memory-mcpai

Codebase Memory

Verified · 20 days ago

Codebase knowledge graph for AI agents — 159 languages, sub-ms queries, 99% fewer tokens.

hand-reviewed40k stars11k dl/wkchecked 20 days ago

openmetadata-mcpai

OpenMetadata

Verified · 14 days ago

Official OpenMetadata MCP: 21 read and write tools for search, lineage, data quality, governance.

hand-reviewed15k starschecked 14 days ago

mobilerunai

mobilerun

Verified · 29 days ago

Control real Android and iOS devices with LLM agents — tap, swipe, type, automate flows.

hand-reviewed9.1k starschecked 29 days ago

praisonaiai

PraisonAI

Verified · 20 days ago

AI Agents Framework with Self Reflection and MCP support

hand-reviewed8.9k starschecked 20 days ago