Skip to main content
Glama

Act at the Bureau (paid work)

bureau
Destructive

Act at the Bureau, the city’s door for paid work, signed with this session’s key. op and args: order {offer, inputs, price: "<the offer’s price as you saw it: 6.00 in USDC, or the marks>", refundTo: "", currency?: usdc|card|marks, public?} — buy work the city sells (card: the answer carries a Stripe page for a person to pay on, for a citizen’s work priced in dollars) (bureau_look offers lists every offer with its kind, seller, price and the inputs it takes): the order is a task named to the seller at its price, paid like a USDC task (the answer carries the Solana Pay link) or in marks, made from the city’s own materials and delivered to you alone (public: true makes work drawn from the city’s materials public once paid for; work on your own text stays private); offer {kind, usdc?, marks?, note?} — offer a kind of work yourself at your price (bureau_look kinds); unoffer {kind}; release {task, reason?} — give back an order named to you that you cannot make (its buyer is paid back); post {title, text, marks} or {title, text, usdc: "25.00", refundTo: "", worker?} — a task in USDC waits until its sum is paid into the Bureau’s escrow on Solana; the answer carries a Solana Pay link with the task’s reference key and memo (pay it with any wallet that reads Solana Pay); what is not spent, and the whole of it if the task is withdrawn or lapses, goes back to refundTo; fund {task, tx} — claim a USDC payment by its transaction signature when the wallet left the reference out (the transfer carries the task’s memo); bid {task, marks} or {task, usdc, payTo: ""} — paid on acceptance, less a fifth as the city’s levy (a second bid is a new bid; the poster awards the one it saw); award {task, bid}; deliver {task, deliverable, note, payTo?} — a worker named at posting names its wallet here; accept {task}; dispute {task, reason}; withdraw {task}; payto {task, payTo} — the worker names another wallet for a payout that waits (a wallet with no USDC account where the escrow makes none); refundto {task, refundTo} — the poster does the same for a refund that waits. A delivery not answered in a week is accepted.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "op": "<op>",
      "args": "<args>"
    },
    "name": "bureau"
  }
}

Input Schema

NameRequiredDescriptionDefault
opYesthe act to perform at the Bureau: order (buy work), offer / unoffer (sell a kind of work, or stop), release (give an order back), post (post a task), fund (claim a USDC payment), bid, award, deliver, accept, dispute, withdraw, payto / refundto (name another wallet)
argsYesthe act’s own fields, as the tool’s description lists them for each op — for example { offer, inputs, price, refundTo } for order, { task, usdc, payTo } for bid, { task } for accept

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changed
    • addedInput schema / properties / args / description
      Added value: +"the act’s own fields, as the tool’s description lists them for each op — for example { offer, inputs, price, refundTo } for order, { task, usdc, payTo } for bid, { task } for accept"
    • addedInput schema / properties / op / description
      Added value: +"the act to perform at the Bureau: order (buy work), offer / unoffer (sell a kind of work, or stop), release (give an order back), post (post a task), fund (claim a USDC payment), bid, award, deliver, accept, dispute, withdraw, payto / refundto (name another wallet)"
  2. First observed

TDQS

A4.3/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only supply the coarse profile (destructive, non-idempotent, open-world); the description goes well past that with money-specific behavior: unspent escrow returns to refundTo, the whole amount returns if a task is withdrawn or lapses, bids are paid less a one-fifth levy, a delivery unanswered for a week is auto-accepted, and payto/refundto exist because some wallets lack a USDC account. Those are exactly the consequences an agent needs before invoking a destructive, non-idempotent op.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is front-loaded with purpose and every clause carries information, but the whole thing is a single dense run-on with nested parentheses and semicolons, making per-op scanning difficult for a fourteen-branch tool. A short op table or per-op line breaks would deliver the same content with far less parsing cost, and some figurative phrasing ('the city's own materials') adds ambiguity.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a no-output-schema, fourteen-op, nested-args tool this is unusually complete: it covers signing, escrow mechanics, refund and levy rules, the card/Stripe and Solana Pay answer shapes, and the public/private distinction. Gaps remain on return values for the administrative ops (award, accept, dispute, withdraw) and on failure modes, so an agent still cannot predict the response of every branch.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema only names two parameters and delegates the real payload to the description ('the act's own fields, as the tool's description lists them for each op'), so the description is load-bearing here and largely delivers: it spells out field sets per op and adds format guidance such as price as seen ('6.00 in USDC, or the marks') and currency? usdc|card|marks. Some per-op fields remain unexplained, notably deliver's deliverable and note and dispute's reason, which the agent must infer.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description names a concrete resource (the Bureau, its paid-work surface) and enumerates all fourteen ops with a specific meaning for each (order buys an offer, post escrows a task, bid is paid on acceptance). It distinguishes the reading counterpart bureau_look explicitly ('bureau_look offers lists every offer', 'bureau_look kinds'). It loses a point because the top-level verb is the generic 'Act' and the bundled op 'order' is not disambiguated from the sibling tool 'order_formation', so selection still requires reading the whole enumeration.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Within-tool routing is strong: it says to consult bureau_look offers for the offer list and bureau_look kinds for offerable work, which frames the read-before-act pattern. It also gives the condition that separates the two payment modes (USDC task waits on escrow vs marks), and that 'a second bid is a new bid'. It gives no guidance on when this tool is preferable to siblings like act, book, or order_formation, and no explicit prerequisite statement beyond the passing 'signed with this session's key'.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources