ANOLISA
ANOLISA (Agentic Nexus Operating Layer & Interface System Architecture) | Agentic OS with runtime, security, observability, and Tokenless response compression f
What is ANOLISA?
ANOLISA (Agentic Nexus Operating Layer & Interface System Architecture) | Agentic OS with runtime, security, observability, and Tokenless response compression for lower token usage and cost.
Overview of ANOLISA
What it is
ANOLISA is an operating system layer for AI Agent workloads, providing runtime, security, observability, and tokenless response compression to reduce token usage and cost. It is a server-side layer that addresses three practical constraints of Agent execution: terminal entry, token cost, and execution environments. The repository includes components such as cosh-ng (an AI-native Linux terminal), Token-less (tool response compression), AgentSight (observability via eBPF on Linux), Agent Sec Core (sandbox and verification), ws-ckpt (checkpoint and rollback), Agent Memory (cross-session memory), SkillFS (focused skill views), and AW (unified Agent policy configuration). The CLI provides a single installation entry point, and each capability can be enabled independently.
Key features
- Token-less response compression reduces token usage before they reach the model, based on measurements showing 65.8% fewer tokens on tool responses, 47.3% fewer tokens on tool schemas, and 62.9% fewer tokens on the full pipeline (results vary by workload).- cosh-ng is an AI-native Linux terminal that keeps familiar Bash/Zsh behavior and adds an Agent that can understand intent, use tools and Skills, and ask for approval before risky work.
- AgentSight uses eBPF on Linux to observe an Agent without changing its code, showing token use and sub-agent branches.
- Agent Sec Core isolates risky operations and detects changes to signed Skills before they run, reporting them as drifted.- ws-ckpt keeps recovery points for workspace changes.
- SkillFS keeps the current skill view focused and makes other skills discoverable when needed.
- Agent Memory reuses useful context across sessions.
Use cases
ANOLISA is intended for AI Agent developers and operators who need to reduce token costs, run Agents directly in the terminal, and have runtime boundaries and security controls for Agent workloads. It can be used with existing Shell, Agent frameworks, and sandbox environments. Specific use cases include running tool-heavy coding tasks with reduced token usage, troubleshooting Agent behavior with eBPF observability, and managing skill integrity with drift detection.
Getting started
bashcurl -fsSL https://get.agentic-os.sh | bashsudo anolisa --install-mode system install cosh-nganolisa install tokenlessAlternatively, to try Token-less with Claude Code:bashcurl -fsSL https://get.agentic-os.sh | bashexport PATH=$HOME/.local/bin:$PATHanolisa install tokenlessanolisa adapter enable tokenless claude-code
License and pricing
ANOLISA is an open-source project released under the Apache License 2.0.
Based on the project's public README and repository metadata.