tokenrip
Live tokenrip agents

Portable agents that remember.

Each agent lives on Tokenrip with its own memory, artifacts, and instructions — then mounts into the runtime where you already work.

Available agents

Real agents you can inspect and mount.

Each agent has a profile with what it does, what it remembers, and how to run it from supported tools.

2 live · updated today

How to call any of these agents

One install, every agent

Install the bootloader once, then call any agent with /tokenrip-bootloader <agent-id> (the slug from the agent’s URL).

Step 1 — one-time install

mkdir -p .claude/commands && curl -fsSL https://api.tokenrip.com/commands/tokenrip-bootloader.md -o .claude/commands/tokenrip-bootloader.md

Installs the tokenrip-bootloader Claude Code slash command. Loads + runs Tokenrip agents end-to-end via the rip CLI. Distinct from the optional tokenrip-cli skill (auto-loaded context for publishing your own work) — install one, both, or neither independently.

Already installed? Skip to step 2.

Step 2 — invoke an agent

/tokenrip-bootloader <agent-id>
§ 1.0
ConceptMemory on Tokenrip.
Surface in your runtime.

The agent lives on Tokenrip. You use it where you work.

A Tokenrip agent keeps its memory, artifacts, and behavior in one durable place. Mounting brings that agent into a supported AI tool or workflow — without starting from scratch.

§ 1.1 Choosestep

Choose an agent.

Browse live agents and open the profile to see what it does, what it remembers, and what it produces.

§ 1.2 Mountstep

Mount it.

Follow the agent's mount instructions for a supported runtime — Claude, Codex, MCP, or your CLI.

§ 1.3 Workstep

Keep working.

The agent builds on memory, references artifacts, and improves across sessions instead of resetting every time.

on tokenrip

Agent identity

MemoryArtifactsInstructionsVersions
mounts into
your runtime

Where you work

ClaudeCodexMCP clientCLI / custom harness

Same agent. Durable memory. Different surfaces.

What makes Tokenrip agents different

Not a prompt. Not a project. An agent with a home.

§ 2.1 Persistent memoryproperty

It remembers across sessions.

Working style, prior decisions, recurring patterns, and open loops carry forward — no re-prompting every time you start.

§ 2.2 Versioned imprintproperty

The instructions are inspectable.

Rules, examples, and behavior are versioned like software. Improve the imprint without losing the memory built on top of it.

§ 2.3 Portableproperty

The same agent moves with you.

One identity. Multiple runtimes. The chat tool is the surface — the agent is the thing that persists.

§ 2.0
PublishFor builders with a
repeatable methodology.

Publish your own agent.

If you have a repeatable methodology and an audience, Tokenrip packages your expertise as a mounted-agent imprint — versioned, memory-capable, portable across runtimes. Your audience pays their own inference; you own the agent.

~/agents/my-imprintzsh
$rip init my-imprint
imprint scaffold created
$rip publish # v0.1
imprint pushed to tokenrip
mount surfaces: claude, chatgpt, cursor, mcp, cli
tokenrip.com/agents/my-imprint
$
Questions

Things people ask about mounted agents.

Mounting brings a Tokenrip agent into a runtime you already use — Claude, ChatGPT, Cursor, an MCP client, or a CLI. The agent’s memory, artifacts, and instructions live on Tokenrip; the runtime is just the surface where you talk to it.
Those tie the agent to a single runtime. A Tokenrip agent has persistent memory across sessions, versioned imprints, and durable artifacts — and the same agent runs across supported tools.
The person running the agent pays their own inference through whichever runtime they use. Tokenrip is not in the inference path — so the agent can reason as deeply as the workflow needs.
Yes. Agents support layered memory — shared imprint and team context, with private context that stays scoped to a single operator.
No. To use an agent, follow the mount instructions on its profile. The CLI is for builders publishing their own imprints.
Claude, ChatGPT, Cursor, MCP-compatible clients, and the Tokenrip CLI. Each agent profile lists its mount instructions per runtime.
§ 3.0
BeginPick one.
Keep the memory.

Pick an agent. Mount it. Keep the memory.

Browse the live agents and open a profile to see what each one remembers and how to bring it into your runtime.

Next

One identity, many runtimes. The agent persists on Tokenrip; the chat tool is just the surface. Mount once, keep the memory across every session that follows.