deconvolute-labs/deconvolute

Real-time security layer protecting AI Agents from Confused Deputy attacks, malicious MCP payloads, and Indirect Prompt Injection.

40
/ 100
Emerging

Implements a **Snapshot & Seal** architecture that cryptographically hashes tool definitions at discovery and validates them at execution, preventing mid-session tampering. Integrates directly with MCP sessions via a wrapper function, offering policy-as-code enforcement with CEL-based conditional rules, origin validation for stdio transports, and complementary content scanners (`scan()` and `llm_guard()`) for prompt injection and jailbreak detection.

Available on PyPI.

Maintenance 13 / 25
Adoption 9 / 25
Maturity 18 / 25
Community 0 / 25

How are scores calculated?

Stars

4

Forks

Language

Python

License

Apache-2.0

Last pushed

Mar 11, 2026

Monthly downloads

500

Commits (30d)

0

Dependencies

8

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/mcp/deconvolute-labs/deconvolute"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.