Use Riley anywhere

Use Riley in Claude Code.

one workspace · same approval gate · same proof

Verified: add Riley with one command and use it from the terminal.

LiveAll surfaces

What it is

Verified: add Riley with one command and use it from the terminal.

What Riley can do here

Ask for work in plain language and Riley does it with the same tools, memory, and rules as everywhere else. Reading and drafting happen freely; anything that writes outside your workspace waits for your approval.

Easy setup

No code required.

Mint a Riley key in Settings (it is shown once).
Run claude mcp add with the endpoint and your key header.
Ask Claude Code to use Riley.

Advanced setup

For technical users.

MCP endpoint (Streamable HTTP)

https://app.hireriley.com/mcp

Auth header

x-api-key: YOUR_RILEY_KEY

Endpoint (Streamable HTTP): https://app.hireriley.com/mcp
Auth header: x-api-key: YOUR_RILEY_KEY (Authorization: Bearer also works)
Tools exposed: riley_audit, riley_respond, riley_social, riley_prospect, riley_memory, riley_health
Keys are per workspace, shown once at mint, and revocable in Settings.

Full protocol notes, the tool list, and supported methods live on the developers page.

Permission model

Claude Code can call Riley's tools with your key; external writes still queue for approval.

Proof captured

Actions taken through this surface appear in your approvals queue and receipts like any other Riley work.

Security

Keys and access.

Keys are per workspace, shown once when you mint them, stored only as a hash, and revocable in Settings at any time. Riley cannot show a key again after minting, so if you lose one, revoke it and mint a new one. No surface gets a side door: every external write still passes through your approvals queue.

See the security page for the full permission model and proof receipts for how work is recorded.