install Temet for Claude Code
Install Temet for your coding agent
Copy one prompt, let your coding agent read your Temet profile, then install the Temet skill and CLI workflow in a lighter, more believable two-step setup.
1. Start from the real profile, not from a heavy integration
The new Temet install flow starts from a public profile page and a machine-readable profile, not from a browser extension, a native app dependency, or a complex runtime check. If a real profile exists, the agent reads that first. If not, Temet falls back to a demo profile and still explains the method clearly.
Install Temet for me.
Temet tracks how I work with AI, extracts the skills I actually demonstrate, and helps me improve over time.
First, read my Temet public profile:
https://temetapp.com/a/{address}
Then, if available, read my machine-readable Temet profile:
https://temetapp.com/api/a/{address}?token={read_token}
Then read the Temet agent method:
https://temetapp.com/agent-docs
Install the Temet skill for this agent:
https://temetapp.com/temet.SKILL.md
copy prompt -> agent reads the profile -> agent joins via HTTP -> local CLI only when needed
2. Step one: copy one prompt into Claude Code, Codex, or another agent
The landing page now does the simplest possible thing: give the user a clear CTA, a prompt they can copy, and machine-readable docs behind it. The agent becomes 'connected' when it actually reads the Temet surfaces over HTTP, not because the site pretends to detect the runtime magically.
Step 1
Connect your agent with one prompt
On the landing page, Temet can now offer a clear CTA: Copy prompt for Claude Code. Copy prompt for Codex. Or paste a neutral prompt into any coding agent. The point is to reduce setup friction at the very beginning.
Step 2
Let the agent install the skill and guide the CLI
Once the agent has read the profile and the Temet method, it can install the Temet skill, explain the current strengths and watchouts, and guide the next real command such as npx @temet/cli audit.
3. Step two: let the agent install the Temet skill and guide the CLI only when needed
After the agent reads the profile and the Temet method, it can install the Temet skill, summarize the strengths and watchouts it sees, and only then guide the user through the next local command such as npx @temet/cli audit. This keeps the setup progressive instead of dumping every step up front. Just run npx @temet/cli audit when the agent has finished reading the profile and installing the skill.
Once the agent has read your profile and the Temet method, this is the next local command it should guide.
Why this install method is smart
It avoids
- Heavy integrations on day one
- Complicated local installs before value is visible
- Runtime detection hacks that break trust
What it gets right
- Do not ask too much setup at the beginning
- Give a clear CTA and a copyable prompt
- Expose machine-readable docs and real HTTP surfaces
- Consider the agent connected only when it actually calls Temet
4. Why this install method is intelligent
It avoids heavy integrations at the beginning, avoids complicated local installs before value is visible, and avoids fragile runtime-detection hacks. The UX feels smooth because the mechanics are actually simple: copy prompt, agent reads Temet over HTTP, then the real workflow begins.
5. What Temet now gets right on the landing page
The landing page can now say something concrete: Connect your agent. Copy prompt for Claude Code. Copy prompt for Codex. Or paste a neutral prompt into any coding agent. That is a much stronger entrypoint than expecting users to read a long setup document before they see the product move.
FAQ
Does Temet automatically detect Claude Code or Codex?
No. The new flow is prompt-based, document-based, and endpoint-based. That is intentional because it is simpler and more honest than pretending to detect every runtime automatically.
What happens if I already have a public Temet profile?
The agent should read your real public profile first, then optionally read the tokenized machine-readable profile, and only then help with the next Temet step.
Why not force the CLI install first every time?
Because the new method reduces cold start. The user can understand Temet through the profile and the prompt first, then run the CLI when the next action actually matters.
Next step
The prompt and skill reduce cold start, but the CLI still matters. The next useful step is to run a local Temet audit on real work and compare it with the profile the agent just read.
Run the local CLI nextPublished March 13, 2026