c2siorg

c2siorg/acf-sdk

The Cognitive Firewall is a Zero Trust control layer for agentic systems that protects the reasoning control plane of large language model (LLM) agents. It prevents prompt injection, context manipulation, memory poisoning, and unsafe tool-output re-injection by enforcing policy-driven validation before any input enters model context.

1 good first / help-wanted issue · Go · last activity Sep 1, 2026

1 Open Issue Need Help Last updated: Sep 1, 2026

Open Issues Need Help

View All on GitHub
c2siorg/acf-sdk
20

The Cognitive Firewall is a Zero Trust control layer for agentic systems that protects the reasoning control plane of large language model (LLM) agents. It prevents prompt injection, context manipulation, memory poisoning, and unsafe tool-output re-injection by enforcing policy-driven validation before any input enters model context.

Go