Quick start: connect Raven

Raven is a TUI-first agent framework — an OpenClaw-style assistant you drive from the terminal (raven agent, raven gateway, channels, cron). Like any agent that can execute commands and speak HTTP, it connects to Plexus in-context: nothing to install, one pasted instruction.

The demo story is the same one we use for every agent: Raven reads your machine's load through Plexus and has Codex generate a "weather report" illustration of it with its image tool — every step granted, audited, and revocable.

What you need:

• A running Plexus gateway (Get running).
• Raven onboarded with any model provider (raven onboard, then raven doctor to verify).
• For the comic finale: the Codex CLI on the gateway machine + the Codex source's Real launch enabled (Console → What I expose → Codex).

1 · Open the console

Open /admin on your gateway and paste your connection key. This is your admin credential — the agent never gets it.

2 · Connect an agent

Connect an agent → name it raven → pick its capability set. Same five as the OpenClaw story:

• sysinfo.resources.read + sysinfo.processes.list
• codex.run
• workspace.list + workspace.read

Checked reads become standing grants. codex.run is execute — per-use by default, and an in-context agent can't idle in an approval loop, so opt it into Standing at connect (double-confirmed):

3 · Hand the instruction to Raven

Pick the In-context / HTTP delivery form and copy the instruction (single-use enroll code baked in):

Paste it into a Raven chat — TUI or one-shot CLI:

raven agent -m "<the instruction you copied>

After you are connected, here is your first task:
1. Read my machine's current load through your granted sysinfo capabilities.
2. Call codex.run ONCE: have Codex use its built-in image generation tool to draw
   a cartoon weather-report illustration of the load, with the real numbers as
   text in the artwork, saved as load-weather.png.
3. Verify the file landed via workspace.list and report its size."

Raven bootstraps itself from the gateway's self-description — enroll → handshake → grant → invoke — and reports back in the terminal:

4 · The payoff

Codex runs headless, write-confined to the workspace, and generates the illustration with its image tool:

5 · What you can see and undo

The console's Activity view has the whole trail — and opening the codex.run invoke gives you the replay locally command that reopens the exact Codex session in your terminal:

The agent's row under Agents shows its standing trust, revocable per grant or wholesale:

Where to go next

• Quick start: OpenClaw — the same loop through a chat-first assistant.
• Connect an agent — all three delivery forms in depth.
• The security model — why reads stand, writes pend, and execute needs your explicit opt-in.