Skip to main content
Glama

propose_contract

Turns an agreed plan into a versioned execution contract with tasks, budget and scope, then collects participant consent before freezing so workers can build in parallel.

Instructions

Turn an agreed plan into a heads-down contract: a draft v1 in a new execution, which workers build in parallel once every participant consents and you freeze it. The structure comes from, in order: contract (an explicit object with tasks: the most reliable route), the last ```roundtable-tasks JSON block in plan or in the archived from_session (an ultraplan started with execution_ready: true ends with one), or a structuring call on the plan text (one retry quoting the validation errors). A contract object without tasks overlays the structure (budget, sandbox, workers, authority, scope.paths, ...). original_prompt is the human's request, verbatim (default: the session's task). repo_dir (default the workspace) resolves to its git top level and base_ref (default HEAD) to a commit. Participants default to the session's agents plus every task persona. consent: true asks each participant for Consent: AGREE | OBJECT (an unclear answer is re-asked once, then counts as an objection). Nothing is written to the repository: the contract lives under ROUNDTABLE_HOME/executions. Returns the contract as markdown and a JSON summary (contract_id, version, status, warnings, consent, next). Show the contract to the human before freezing.

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": {
      "goal": "<goal>",
      "orchestrator": "<orchestrator>"
    },
    "name": "propose_contract"
  }
}

Input Schema

NameRequiredDescriptionDefault
goalYesThe ultimate goal the contract delivers.
planNoPlan text: its last roundtable-tasks block is used, else a structuring call turns it into tasks.
modelNoModel id. Defaults: claude-sonnet-5-5 (anthropic), gpt-5.6-luna (openai), zai-org/GLM-5.3 (together), qwen/qwen3-235b-a22b-instruct-2507 (replicate), qwen3:14b (ollama), default (openai_compatible). Model aliases: "opus" (claude-opus-5-5), "sonnet" (claude-sonnet-5-5), "haiku" (claude-haiku-5-5). Replicate accepts owner/name, owner/name:version, or a https://replicate.com/owner/name URL.
titleNoShort title; default the goal's first line.
consentNoDefault false. True: ask every participant for consent right away.
base_refNoThe commit or ref the run starts from; default HEAD. Stored as the resolved sha.
base_urlNoOnly with provider "openai_compatible": base URL of an OpenAI-compatible /v1 endpoint (e.g. http://localhost:1234/v1). Overrides OPENAI_COMPATIBLE_BASE_URL. OPENAI_COMPATIBLE_API_KEY, if set, is sent only to the OPENAI_COMPATIBLE_BASE_URL origin or one listed in ROUNDTABLE_OPENAI_COMPATIBLE_TRUSTED_ORIGINS; any other base URL gets no Authorization header and must be https (http only to loopback, or to private addresses with ROUNDTABLE_ALLOW_PRIVATE_BASE_URLS=1), never a cloud metadata or (without that setting) private address.
contractNoAn explicit contract (roundtable.contract/1 fields: title, scope.paths, participants, teams, tasks, interfaces, decisions, dissents, validation {commands, final}, budget, authority, sandbox, workers, connectors). id, version, status and consent are set for you.
providerNoLLM provider. Default: anthropic. Options: anthropic, openai, together, replicate, ollama (local, when reachable), openai_compatible (set OPENAI_COMPATIBLE_BASE_URL or pass base_url).
repo_dirNoA directory inside the repository to build in; default ROUNDTABLE_WORKSPACE_DIR.
artifactsNoRepository paths or URLs the goal refers to.
draft_planNoDuel from_session only, default true: when neither contract (with tasks) nor plan is given, draft a task plan from the winning artifact, the judges' reasons and the dissents (deterministic, no model call). False imports only the decision and dissents, and then contract or plan is required. A tie, no winner or no artifact drafts no plan.
parent_taskNoFor a child contract: the parent's subcontract task this contract carries out.
from_sessionNoId of an archived session (an ultraplan, collaboration or meeting that has ended) whose roundtable-tasks block or plan to build from, or of an archived duel (duel-<id>): its winning artifact title becomes a decision and its dissents become draft concerns (a tie or no winner imports no decision). A duel has no task list, so by default a deterministic task plan is drafted from the winning artifact (see draft_plan); an explicit contract (with tasks) or plan replaces it.
orchestratorYesYour name as this contract's orchestrator (e.g. main). Only the orchestrator that proposed a contract can amend or freeze it.
participantsNoAgents who must consent (names or .md paths); task personas are always added.
original_promptNoThe human's request, verbatim (default: from_session's task or topic). Workers see it on every task card.
parent_contractNoFor a child contract: the parent contract id (with parent_task, a subcontract task of it).

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv2.0.0

TDQS

A4.2/5.0
Behavior4/5

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

Annotations declare a non-readonly, non-idempotent, open-world write, and the description adds real context beyond that: nothing is written to the repository, the contract lives under ROUNDTABLE_HOME/executions, consent is solicited per participant with one re-ask before an objection counts, and structuring calls retry once quoting validation errors. It still does not say what happens on a failed consent round or how long consent waits.

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?

Purpose is correctly front-loaded and there is essentially no filler, but the body is one long run-on paragraph built from stacked parentheticals (task source order, consent rules, defaults) that forces re-reading. Splitting the source-resolution order and consent behavior into separate sentences would cost nothing and gain much.

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 an 18-parameter tool with a nested contract object and no output schema, the description covers the important ground: where structure comes from and in what order, what overlays vs. what replaces, defaults, side effects, and the returned markdown plus JSON summary keys. It is slightly thin on failure paths (invalid contract, unreachable provider) and on the child-contract parameters.

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

Parameters4/5

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

Schema coverage is 100%, so the schema baseline is already met; the description nonetheless adds resolution semantics the schema does not state, such as repo_dir resolving to the git top level, base_ref defaulting to HEAD resolved to a sha, participants defaulting to session agents plus task personas, and a contract object without tasks acting as an overlay. It omits the ordering interplay among parent_contract/parent_task and from_session beyond a brief mention.

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

Purpose5/5

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

States a specific verb and product: 'Turn an agreed plan into a heads-down contract: a draft v1 in a new execution.' It distinguishes this creation step from the later freeze step ('once every participant consents and you freeze it'), so an agent can tell it apart from amend_contract/freeze_contract.

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?

The opening clause ('an agreed plan') sets the precondition for use, and the closing line routes the agent to the next step ('Show the contract to the human before freezing'). It does not explicitly name when to prefer amend_contract over re-proposing, so the exclusion side is only implied.

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