Apache-2.0 Agentic OS layer (Rust) for agent workloads; Token-less plugs between any agent framework and the model with no code changes, so it works alongside DeepSeek Harness setups.
DSH integration
Compatible
Inferred
Safety audit
Unaudited
Last verified
2026-08-21
License
Apache-2.0
01What can it help you accomplish?
Cut token cost of tool-heavy agent workflows before responses reach the model
Token-less response/schema compression in the pipeline — README reports 65.8% fewer Tokens for tool responses, 47.3% for tool schemas and 62.9% for the full pipeline, with dropped items retrievable via reversible `<<tokenless:KEY>>` markers
Developers running token-heavy agent tasks (e.g. alongside DeepSeek Harness workflows) who want lower context cost without changing their agent framework
Let an Agent operate the system directly from the terminal
cosh-ng AI-native terminal that keeps Bash/Zsh behavior and adds an Agent which understands intent, uses tools and Skills, and asks for approval before risky work
Linux/macOS users and DevOps engineers who want agent-driven system work in one terminal instead of a separate chat app
Secure agent execution with sandboxing, Skill integrity checks and workspace recovery
Agent Sec Core isolates risky operations and flags changed signed Skills as `drifted`/`deny`; ws-ckpt keeps recovery points for workspace changes
Teams running unattended agent tasks who need execution boundaries, Skill tamper detection and a way back
02How to install into DeepSeek Harness
Prerequisites
- Linux or macOS (per the README platform badge)
- curl plus shell access; the bootstrap script installs into the user environment (`export PATH="$HOME/.local/bin:$PATH"`)
- sudo for system-mode installs such as cosh-ng
Installation steps
- 01
Install the ANOLISA CLI entry point: `curl -fsSL https://get.agentic-os.sh | bash` (then `export PATH="$HOME/.local/bin:$PATH"`)
$ curl -fsSL https://get.agentic-os.sh | bash
- 02
Install the AI-native terminal in system mode: `sudo anolisa --install-mode system install cosh-ng`
- 03
Add Token-less independently: `anolisa install tokenless`
- 04
For Claude Code, connect the adapter: `anolisa adapter enable tokenless claude-code`, then restart Claude Code
- 05
Run `cosh` to enter the AI-native terminal
Verify the integration
Not specified by the author
03DSH integration and capability boundaries
Framework-agnostic companion layer: Token-less compresses tool responses/schemas between any Agent framework and the model with no harness changes, so it can sit alongside DeepSeek Harness workflows; the README does not mention dsh by name
Token-less tool-output compression
tool schemas and tool responses between the Agent and the model→compressed context with field-blacklist drops; dropped array items stay retrievable through reversible `<<tokenless:KEY>>` markers; `tokenless stats summary` / `tokenless stats list` show savings
runs as a pipeline between the Agent and the model, so it intercepts tool traffic in flightadapter configuration writes into the connected agent (e.g. Claude Code) and requires restarting itcosh-ng AI-native terminal
natural language and shell commands in one terminal→an Agent that understands intent, uses tools and Skills, and asks for approval before risky work, while keeping Bash/Zsh behavior
installed in system mode via `sudo anolisa --install-mode system install cosh-ng`Agent Sec Core sandbox & Skill verification
agent operations and signed Skills→risky operations isolated in a sandbox; changed signed Skills reported as `drifted` and blocking findings recorded as `deny` on rescan
isolates/intercepts risky operations at runtime and records security findings for SkillsAgentSight observability
a running Agent on Linux→end-to-end view of user input through model and tool calls, with Token use and sub-agent branches, observed via eBPF without changing agent code
uses eBPF kernel-level tracing on Linux to observe the agent process
04Who is it for? When not to use it?
Good for
- Developers running token-heavy agent tasks (e.g. alongside DeepSeek Harness workflows) who want lower context cost without changing their agent framework
- Linux/macOS users and DevOps engineers who want agent-driven system work in one terminal instead of a separate chat app
- Teams running unattended agent tasks who need execution boundaries, Skill tamper detection and a way back
Not for
- ANOLISA targets Linux and macOS only (per the README platform badge); several runtime features such as AgentSight's eBPF observation are Linux-specific.
- Installation uses a remote bootstrap script (`curl -fsSL https://get.agentic-os.sh | bash`), requires network access, and system-mode components like cosh-ng need sudo.
05Compatibility, maintenance and safety notes
- ANOLISA targets Linux and macOS only (per the README platform badge); several runtime features such as AgentSight's eBPF observation are Linux-specific.
- Installation uses a remote bootstrap script (`curl -fsSL https://get.agentic-os.sh | bash`), requires network access, and system-mode components like cosh-ng need sudo.
- Token savings vary by workload: the cited 317K Tokens (40.5%) came from one observed coding task, and savings apply to the tool responses entering the context, not to the whole session bill.
Apache-2.0 · actively maintained (release tokenless/v0.7.11, 2026-08-20; repo pushed 2026-08-19)
06Frequently asked questions
How does ANOLISA integrate with DeepSeek Harness?
The README does not mention DeepSeek Harness (dsh) by name. ANOLISA is framework-agnostic: Token-less runs as compression between the Agent and the model — "no Agent framework code changes" — so it can optimize tool calls from an existing agent setup, and its adapters connect to specific agents (the README shows Claude Code).
Which platforms are supported?
Linux and macOS, per the README platform badge. Some capabilities are Linux-only, e.g. AgentSight's eBPF-based observation.
What are the install prerequisites?
Network access and curl to run the bootstrap script `curl -fsSL https://get.agentic-os.sh | bash`, a PATH export to `~/.local/bin`, and sudo for system-mode installs like cosh-ng.
How much token cost can Token-less save?
Benchmarks in the README show 65.8% fewer tokens for tool responses, 47.3% for tool schemas and 62.9% for the full pipeline; one observed coding task saved 317K tokens (40.5%). Results vary by workload, and savings apply to tool responses entering the context, not the whole session bill.
Is the compression safe and reversible?
Yes — dropped array items stay retrievable through a `<<tokenless:KEY>>` marker, keeping the compression reversible, and ANOLISA keeps the Shell, agent framework and sandbox you already run.
07Related DSH workflows
weknora
by tencent
Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki.
honcho
by plastic-labs
Memory library for building stateful agents
mirage
by strukto-ai
The World's First Virtual Terminal for AI Agents
reme
by agentscope-ai
ReMe: Memory Management Kit for Agents - Remember Me, Refine Me.
08Data and sources
The operating system layer for Agent workloads. Let Agents drive the system straight from your terminal, and strip the t…
Compression runs between the Agent and the model, so no Agent framework code changes.
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-21. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.