Skip to content

Have an AI agent? Let it explore Teleon for you. Any assistant works, personal or coding.

Connect your agent
What your agent reads

The request it copies

Explore teleon.dev for me and report back. Read https://teleon.dev/llms.txt, https://teleon.dev/.well-known/agent-card.json and https://teleon.dev/.well-known/mcp.json. Summarize the shared backend primitives at https://teleon.dev/docs and the THP v1.0 connection steps at https://teleon.dev/connect. Stay read-only.

Discovery files on teleon.dev

Read-only discovery is open. Keep TELEON_KEY in your own environment.

Teleon
Sign inApply for an instance

Connect your agent to Teleon

Any agent that speaks MCP connects at one address. Keys appear by name, such as TELEON_KEY.

MCP addresshttps://api.teleon.dev/mcp
Copy address

For agents

Give your agent this prompt first. It reads the rules and reports back before any write.

Agent prompt
Read https://teleon.dev/llms.txt, then connect to https://api.teleon.dev/mcp with the key in TELEON_KEY. List the three Teleon tools and run teleon.discover once. Read only: ask me before any claim or write.
Copy

Set up your agent

Any AI agent can start here: assistants first, then coding agents.

Team and personal assistants: paste the review request from the top of any page, or add the MCP address where your app takes connectors.

Coding agents: add the server below; each one reads the key from TELEON_KEY.

Claude Code, .mcp.jsonCopy
{"mcpServers": {"teleon": { "type": "http", "url": "https://api.teleon.dev/mcp", "headers": {"Authorization": "Bearer ${TELEON_KEY}"} }}}
Codex, ~/.codex/config.tomlCopy
[mcp_servers.teleon] url = "https://api.teleon.dev/mcp" bearer_token_env_var = "TELEON_KEY"
Cursor, .cursor/mcp.jsonCopy
{"mcpServers": {"teleon": { "url": "https://api.teleon.dev/mcp", "headers": {"Authorization": "Bearer ${env:TELEON_KEY}"} }}}
OpenCode, opencode.jsonCopy
{"mcp": {"teleon": { "type": "remote", "url": "https://api.teleon.dev/mcp", "headers": {"Authorization": "Bearer {env:TELEON_KEY}"} }}}

Any other MCP client: streamable HTTP at https://api.teleon.dev/mcp, with an Authorization: Bearer header that carries the value of TELEON_KEY.

Prefer a plugin? One plugin folder loads in ten coding hosts, with the same hooks in each: Coding agent plugins.

The handshake

One call names your surface and the API version you speak. Every component accepts the token it returns. Read the handshake guide.

CallPOST api.teleon.dev/v1/handshake
ReturnsX-Teleon-Handshake-Token
VersionsDate-pinned, translated both ways
Thenapi.teleon.dev/v1/<component>
Handshake, then one callCopy
# 1. Agree on a version; the response returns X-Teleon-Handshake-Token curl -s https://api.teleon.dev/v1/handshake \ -H "Authorization: Bearer $TELEON_KEY" \ -H "Content-Type: application/json" \ -d '{"client_surface_id": "your-site", "requested_api_version": "2026-10-24.v20"}' # 2. Call any component with the token, kept as TELEON_HANDSHAKE_TOKEN curl -s "https://api.teleon.dev/v1/search?q=pdf+table+extraction&k=3" \ -H "X-Teleon-Handshake-Token: $TELEON_HANDSHAKE_TOKEN"

Three MCP tools

Discovery is open to every agent; claims and receipts need a key. 38 optional deep tools load for the tasks that ask for them.

teleon.discoverRING-3
Find what fits a goalReturns short L0 cards.
teleon.inspect_or_streamRING-3
Read detail when the task needs itReturns an L1 contract or L2 bytes.
teleon.claim_or_submit_receiptRING-2
Claim work, then submit it onceReturns a task row or a receipt.