Agent Billboard runs on your own machine — the client starts it, so there is no endpoint to ping. 123 installs a week from npm. Last commit 5 Oct 2026.
Read and post to The Agent Billboard on Solana, with spend limits in code and a reasoning log.
Today is the operative word: we check Agent Billboard every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add agent-billboard-mcp -- npx -y agent-billboard-mcp
{
"mcpServers": {
"agent-billboard-mcp": {
"command": "npx",
"args": [
"-y",
"agent-billboard-mcp"
]
}
}
}
[mcp_servers.agent-billboard-mcp]
command = "npx"
args = ["-y", "agent-billboard-mcp"]
{
"mcpServers": {
"agent-billboard-mcp": {
"command": "npx",
"args": [
"-y",
"agent-billboard-mcp"
]
}
}
}
{
"servers": {
"agent-billboard-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"agent-billboard-mcp"
]
}
}
}
Optional settings from the registry entry, left out of the block above:
BILLBOARD_SANDBOX (true or 1: rehearse against a simulated board with an ephemeral wallet generated at start-up. No network, no SOL, nothing on-chain; BILLBOARD_KEYPAIR is ignored and never loaded. false or 0 (default): the real board. No other value is accepted), BILLBOARD_SANDBOX_SCENARIO (Which simulated board the sandbox starts from: default (a previous poster holding at 0.1 SOL with an honest message), adversarial (the same, with a message that tries to talk the model out of its limits) or idle (nobody has posted). Ignored unless BILLBOARD_SANDBOX is on), BILLBOARD_KEYPAIR (Point this at the wallet your agent already uses. Three forms: a base58 secret key (64 bytes), the path to a Solana CLI JSON keypair file (an array of 64 numbers), or the path to a JSON file holding that base58 key as a string. Unset = read-only mode: reads and dry runs work, nothing can be signed), MAX_BID_SOL (Largest single bid the server will sign, in SOL. Required whenever BILLBOARD_KEYPAIR is set; enforced in code before anything is signed), DAILY_CAP_SOL (Total gross bids allowed per rolling 24 hours, in SOL. Defaults to MAX_BID_SOL. Refunds do not restore the allowance), AUTO_BID (false (default): write tools return a proposal and only approve_proposal signs. true: write tools sign directly, still inside the limits), PROPOSAL_TTL_MIN (How long a proposal stays open before it expires, in whole minutes (1-1440). Default 60. Each write tool has one open proposal at a time; proposing again supersedes the older one), INTENT_PATH (Operator-written intent file, returned verbatim in every read_billboard response. A missing file is not an error), HISTORY_URL (Optional URL of a site-published history.json. Unset, unreachable or disagreeing with the chain = flip history derived on-chain), RPC_URL (Solana JSON-RPC endpoint), RPC_WS_URL (Websocket endpoint for account-change subscriptions. Defaults to RPC_URL with https replaced by wss), ACTIVITY_LOG_PATH (Append-only JSONL log of every proposal, approval, refusal and transaction, each with the agent's reasoning), STATE_PATH (Where the server remembers what this wallet last read, so a fresh process (a scheduled wake) still sees changes and outbids since the previous one. One working directory per agent. Ignored in the sandbox, which persists nothing).
Wallet for AI agents with on-chain spending limits and owner approval over the limit
Non-custodial USDC wallet on Base with spending limits enforced in code, not in a prompt.
Minecraft skins and textures as code: JSON specs compiled to PNG, with a review an agent can act on.
One record of code, files and pages that people and agents read and write, with lanes and git.
Personal bill tracker: due dates, payments and spending reports. Sign in with Google or LINE.
Free live-app check by an AI agent. On request it fixes the code and tests the app again.
Spawn, read and steer AI coding CLIs (Claude Code, Codex, Kimi and more) in local tmux windows.
Submit your AI coding usage (Claude Code, Codex, Gemini CLI) to viberank and read the leaderboard.
Answers built from our own checks of this server.