turbo-cloud
ExducerAI GPU cloud for coding agents
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/exducer-ai--exducer-agent-kit)HTML
<a href="/listings/exducer-ai--exducer-agent-kit"><img src="/api/listings/exducer-ai--exducer-agent-kit/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Paste this into Claude Code, Codex, Cursor or another coding agent:
Read https://exducer.ai/agents.md and set up ExducerAI.
The agent installs the CLI, asks you to create an API key and to type it into `turbo login` yourself, then installs the skills and registers the MCP server. The steps it follows are in [SETUP.md](SETUP.md).
Start with a **sandbox key** (https://cloud.exducer.ai, Account, then API keys, with the Sandbox key option). It runs the same commands against a simulated lifecycle, so the agent can rehearse a booking from request to stop and nothing is charged.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 27 lines
- {
- "name": "turbo-cloud",
- "owner": {
- "name": "ExducerAI",
- "email": "hello@exducer.ai"
- },
- "metadata": {
- "description": "ExducerAI for coding agents: skills and the MCP server for renting GPUs in Montréal and calling the Models API.",
- "version": "1.0.0"
- },
- "plugins": [
- {
- "name": "turbo",
- "source": "./plugins/turbo",
- "description": "Rent whole GPUs on ExducerAI from your coding agent: check availability, quote, book, wait, connect over SSH, stop, and read billing. Six skills and the MCP server.",
- "category": "infrastructure",
- "keywords": [
- "gpu",
- "cloud",
- "mcp",
- "training",
- "inference",
- "canada"
- ]
- }
- ]
- }
README.md · README · 143 lines · 2 findings marked
- # ExducerAI GPU cloud for coding agents
- ExducerAI rents whole GPUs in Montréal by the hour and serves an OpenAI-compatible Models API.
- This repository is the kit that lets a coding agent use it for you: check availability, quote,
- book, wait, connect over SSH, stop, and read billing, with your approval at every step that costs
- money.
- | | |
- | --- | --- |
- | MCP server | `https://exducer.ai/mcp` (streamable HTTP), `ai.exducer/turbo` in the MCP Registry |
- | CLI | `turbo`, one Python file: `curl -sSL https://exducer.ai/cli/install.sh \| sh` |
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 11 on GitHub
- | Skills | Six agent skills, also packaged as a Claude Code plugin |
- | Documentation | https://exducer.ai/docs, and https://exducer.ai/llms.txt for agents |
- The MCP server and the API are hosted services. Their code is not in this repository. The setup
- text, the skills and the CLI here are copies of what [exducer.ai](https://exducer.ai/agents) serves;
- if a copy and the site ever differ, the site is right (`sh scripts/check-against-site.sh` compares).
- ## Set up with one prompt
- Paste this into Claude Code, Codex, Cursor or another coding agent:
- ```
- Read https://exducer.ai/agents.md and set up ExducerAI.
- ```
- The agent installs the CLI, asks you to create an API key and to type it into `turbo login`
- yourself, then installs the skills and registers the MCP server. The steps it follows are in
- [SETUP.md](SETUP.md).
- Start with a **sandbox key** (https://cloud.exducer.ai, Account, then API keys, with the Sandbox
- key option). It runs the same commands against a simulated lifecycle, so the agent can rehearse a
- booking from request to stop and nothing is charged.
- ## Set up by hand
- ```bash
- curl -sSL https://exducer.ai/cli/install.sh | sh # one Python 3 file, no dependencies
- Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
- Line 38 on GitHub
- turbo login # type the key here, not into a chat
- turbo agent setup # skills and MCP server, for the agents on this machine
- turbo agent status # what is installed where
- ```
- `turbo agent setup --dry-run` prints the changes without making them, and `turbo agent remove`
- undoes them. More in [cli/README.md](cli/README.md).
- In Claude Code the skills and the server also install as a plugin. Use it instead of
- `turbo agent setup`, not as well as it:
- ```bash
- claude plugin marketplace add https://exducer.ai/agents/marketplace.json
- claude plugin install turbo@turbo-cloud
- ```
- ## The MCP server only
- No key is needed for prices, availability, the catalog, quotes and the documentation:
- ```bash
- claude mcp add --transport http turbo https://exducer.ai/mcp
- ```
- With a key, the account tools answer too:
- ```bash
- claude mcp add --transport http --scope user turbo https://exducer.ai/mcp \
- --header "Authorization: Bearer $TURBO_API_KEY"
- ```
- Other agents and the stdio route through the CLI are in
- [examples/mcp/clients.md](examples/mcp/clients.md).
- ## What the agent may do without you
- These rules are enforced by the server and the CLI, so they hold whatever the agent was told.
- - Reading is open within the key's scope: availability, prices, quotes, your deployments and their
- events, the cost so far, your balance and usage.
- - `create_booking`, `extend_deployment`, `set_auto_renew` and `stop_deployment` refuse unless the
- call carries `confirm: true`. The skills tell the agent to set it only after you approved the
- quote or, for a stop, after you were told the disk is deleted.
- - Every booking and extension carries an idempotency key, so a retried call replays the first
- answer instead of booking twice.
- - API keys, SSH keys, storage keys and webhooks cannot be changed through the MCP server. They stay
- with you, in the console.
- A read key is enough for an agent that only reports. Give it a write key when it should book and
- stop. It never needs an admin key.
- ## Tools
- The server has 22 tools. The "Key" column is the least key a tool needs.
- | Tool | Key | What it does |
- | --- | --- | --- |
- | `get_pricing` | None | List prices in CAD: GPU-hour per GPU model, object storage per GB-month, and the Models API. |
- | `check_availability` | None | One yes or no per GPU model: is it free right now. |
- | `get_service_status` | None | The public status of each service (console, API, GPU region, object storage, Models API). |
- | `list_servers` | None | The servers a booking can name: id (the server_id of quote_booking and create_booking), GPU model, price per GPU-hour in CAD, allowed_gpu_counts, region, and whether it is free now. |
- | `list_templates` | None | The templates a booking can name: slug, name, the container image it runs, category. |
- | `quote_booking` | None | What a booking would cost in CAD: GPU cost, tax and total for the whole duration, and the hourly rates. |
- | `search_docs` | None | Search the documentation by keyword. |
- | `read_doc` | None | One documentation page as Markdown, by its slug (from search_docs). |
- | `get_account` | Read | The account behind the API key (email, company, type), whether the key is a sandbox key, and the prepaid balance in CAD (simulated under a sandbox key). |
- | `list_deployments` | Read | Every deployment of the account with its state, GPU count, rate, cost so far and end time. |
- | `get_deployment` | Read | One deployment in full: state and reason, SSH host and port, the host key fingerprint to verify, rate per GPU-hour, cost so far, start and end time, endpoints. |
- | `get_deployment_events` | Read | The state changes of a deployment, oldest first, each with its reason, actor and time. |
- | `wait_for_deployment` | Read | Poll a deployment until it reaches a state (default running) or a final state, for at most 50 seconds per call. |
- | `get_billing` | Read | The prepaid balance and the most recent credit ledger lines (top-ups and charges) in CAD. |
- | `get_usage` | Read | What the account was charged for GPU time, object storage and the Models API over a date range, in CAD before tax. |
- | `list_invoices` | Read | The account's invoices (not available to a sandbox key). |
- | `list_ssh_keys` | Read | The SSH public keys saved on the account: a booking with no ssh_public_key uses them (not available to a sandbox key). |
- | `get_storage_usage` | Read | Object storage: GB stored, active keys, price per GB-month and the amount billed this month (not available to a sandbox key). |
- | `create_booking` | Write | Send a GPU request. |
- | `extend_deployment` | Write | Add hours to the term of a deployment that is requested, queued, provisioning or running. |
- | `set_auto_renew` | Write | Turn automatic renewal of a deployment on or off. |
- | `stop_deployment` | Write | Stop a deployment now: billing ends at this minute and its container and NVMe scratch disk are deleted and cannot be recovered. |
- ## Examples
- [examples/](examples/README.md) has small, complete examples: the public API and the MCP server
- with no key at all, webhook signature checks in Python and Node.js, a simulated booking on the
- sandbox with the CLI and with the REST API, and the Models API.
- ## In this repository
- | Path | What it is |
- | --- | --- |
- | [`SETUP.md`](SETUP.md) | The setup an agent follows, as served at https://exducer.ai/agents.md |
- | [`plugins/turbo/skills/`](plugins/turbo/skills) | The six skills, as listed at https://exducer.ai/.well-known/agent-skills/index.json |
- | [`plugins/turbo/`](plugins/turbo), [`.claude-plugin/`](.claude-plugin) | The Claude Code plugin and its catalog |
- | [`cli/`](cli/README.md) | The `turbo` CLI, as served at https://exducer.ai/cli/turbo.py |
- | [`examples/`](examples/README.md) | Examples for the API, the MCP server, webhooks, the CLI and the Models API |
- | [`server.json`](server.json) | The listing published to the MCP Registry |
- | [`scripts/check-against-site.sh`](scripts/check-against-site.sh) | Compares the copies here with the site |
- | [`SECURITY.md`](SECURITY.md), [`CHANGELOG.md`](CHANGELOG.md) | How to report a security problem; what changed |
- ## Help
- - Documentation: https://exducer.ai/docs
- - Service status: https://exducer.ai/status
- - Questions: hello@exducer.ai
- - Security: security@exducer.ai ([SECURITY.md](SECURITY.md))
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.