Optimization that fits your development workflow.

Run Artemis from your editor, during code review, before a release, or on a schedule. Every run ends with a pull request containing the best-performing version and benchmark results.

Let your coding agent write the code.Let Artemis optimize it.

When a coding agent reaches an optimization task, it hands it to Artemis. Artemis explores multiple solutions, benchmarks them on your workload, and returns the best-performing one. Continue working in your terminal while Artemis searches in the background.

✳︎Claude Code

Press Shift Tab to automatically approve code edits

ctrl esc to focus or unfocus Claude

Remote ControlSonnet 5

Auto

Connect Your Agent

The Artemis CLI lets a local coding agent drive the platform, straight from your terminal.

Quickstart

Copy this into your coding agent

Your agent will set up the CLI and skills. Optionally it can set up a runner and run a demo.

Set up Artemis on this machine with me.

Artemis deployment: https://artemis.turintech.ai
Skills and setup: https://docs.artemis.turintech.ai/features/artemis-agent-skills
CLI and authentication: https://docs.artemis.turintech.ai/features/artemis-cli
API keys: https://docs.artemis.turintech.ai/settings/user-settings#api-keys
Runners: https://docs.artemis.turintech.ai/features/artemis-custom-runner
Optional demo: https://docs.artemis.turintech.ai/optimization/discover/tutorials/particle-life-example

1. Identify whether you are running in Claude Code, Cursor, Codex, or GitHub Copilot.

Install or update the official Artemis Agent Skills from https://github.com/turintech/skills at user scope using the supported method for this host.

Use only that official repository. Inspect what will be installed before overwriting existing skills, and do not execute scripts from the repository during skill installation.

Prefer `gh skill install turintech/skills --all --agent <host> --scope user` when the installed GitHub CLI supports `gh skill`.

For Claude Code, the TurinTech plugin marketplace is also supported.

If `gh skill` is unavailable, use the host's documented user-level Agent Skills directory and install only the directories under `plugins/artemis/skills`.

If installation requires a reload or restart, tell me exactly what to do and stop until I return.

Then confirm that the `artemis`, `cli-setup`, and `runner-setup` skills are available.

2. Use the `artemis` router and `cli-setup` skill.

Inspect the existing CLI before changing it. Install or update it only if needed, configure it for the deployment above, and verify the installed version and `artemis status`.

Never ask me to paste an API key, registration token, password, or other secret into chat.

If authentication is needed, point me at the public API keys docs above so I can create a key in my own account, then have me enter it interactively in my own terminal.

3. Ask what I want to do with Artemis if it is not already clear.

Maintain does not require a runner. Discovery and Validation do.

If a suitable runner is already online, reuse it.

If one is required and missing, use `runner-setup` and the public runner docs.

First check that this machine is safe for repository code and has the required toolchains, disk, memory, and network access.

Then offer to install, register, and start the runner for me. Get my explicit permission before starting the long-lived process.

If `tmux` is installed, prefer a named tmux session so I can see and manage the runner, and tell me how to attach, detach, and stop it.

Otherwise use a visible terminal or clearly report the background process and how to stop it.

After I agree, start and verify the runner.

Do not put an API key or registration token on a command line where it may be exposed in shell history or process listings.

Ask before creating an operating-system service.

If I want a demo, follow the Particle Life tutorial after setup is ready:
https://docs.artemis.turintech.ai/optimization/discover/tutorials/particle-life-example

4. Finish with a readiness report:

- agent host and skills verification
- CLI version, deployment, authenticated user, and status
- runner reused, installed, or intentionally skipped
- any remaining human action

Do not claim success without verification evidence.

Turn every code changeinto a better one.

Artemis reviews the diff the way a performance engineer would, then returns a branch with the optimization already applied and benchmark results attached.

speech-stack/Changeset

File changesValidations

Keep attention on the fast kernelPR openPull request

Ref:main←Compare:3f2a9c11 file changed+2 −0

whisper/

model.py+2 −0

whisper/model.py@@ -14,6 +14,8 @@

14def cross_attn(q, k, v, mask=None):

15+ k = k.repeat(q.shape[0], 1, 1)

16+ v = v.repeat(q.shape[0], 1, 1)

17 return F.scaled_dot_product_attention(

18 q, k, v, attn_mask=mask

19 )

Scan the whole codebase,not just the diff.

Artemis scans every module in your codebase, validates each optimization on your workload, and opens a separate pull request for every improvement that performs better.