delimit

Official

Overview Schema Source Related Servers Discussions Admin

Delimit Deliberate

delimit_deliberate

Run an AI panel to deliberate and reach consensus on critical decisions. Detect cross-model contradictions for pricing, naming, and strategic choices.

Instructions

Run multi-model consensus via AI-to-AI deliberation (Pro).

When to use: for foundational decisions (pricing, naming, public-facing copy framing, doctrine edits), external PR diffs, or any decision where cross-model contradiction-detection adds value. When NOT to use: for routine implementation choices (orchestrate in-thread or via subagent dispatch) — deliberation is for cross-checked confabulation, not capability.

Sibling contrast: delimit_models manages which providers can be called; this runs the actual panel. delimit_security_deliberate is the security-class variant.

Side effects: writes transcripts under save_path when provided. Models are called via configured providers; Free tier uses 3 builtin slots, Pro/Premium uses BYOK from ~/.delimit/models.json. Strategic / social scopes enforce a 3-model minimum and may invoke Grok as a tiebreaker.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`mode`	No	"dialogue" (short turns) or "debate" (long essays). Default "dialogue".	dialogue
`scope`	No	Optional scope override — "strategic", "social", or "operational". Empty = engine classifies from keywords.
`context`	No	Background context shared to all models.
`question`	Yes	The question to reach consensus on. Required.
`save_path`	No	Optional file path to save the full transcript.
`max_rounds`	No	Max rounds. Default 3 for debate, 6 for dialogue.
`context_files`	No	Optional list of file paths whose contents are read server-side, redacted (secrets/PII), size-capped, and injected as a "Referenced Files" block so the panel can reason over real source (panelists have no filesystem access). Fails closed per file.

Output Schema

TableJSON Schema

Name	Required	Description	Default
No arguments

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedv4.13.2Sep 5, 2026

addedInput schema / properties / context_files

Added value: +{
+  "anyOf": [
+    {
+      "items": {
+        "type": "string"
+      },
+      "type": "array"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Optional list of file paths whose contents are read server-side, redacted (secrets/PII), size-capped, and injected as a \"Referenced Files\" block so the panel can reason over real source (panelists have no filesystem access). Fails closed per file."
+}

Changed5 schema fields changedv4.7.9Jun 9, 2026

changedInput schema / properties / context / description

Previous value: -"Background context for all models."New value: +"Background context shared to all models."

changedInput schema / properties / max_rounds / description

Previous value: -"Maximum rounds (default 3 for debate, 6 for dialogue)."New value: +"Max rounds. Default 3 for debate, 6 for dialogue."

changedInput schema / properties / mode / description

Previous value: -"\"dialogue\" (short turns) or \"debate\" (long essays)."New value: +"\"dialogue\" (short turns) or \"debate\" (long essays). Default \"dialogue\"."

changedInput schema / properties / question / description

Previous value: -"The question to reach consensus on."New value: +"The question to reach consensus on. Required."

changedInput schema / properties / scope / description

Previous value: -"Optional scope override — \"strategic\", \"social\", or\n\"operational\". When empty, the engine classifies from\nkeywords in the question and context. Strategic and social\nscopes enforce the 3-model minimum (charter consensus-thresholds)\nand allow Grok as a tiebreaker on deadlock."New value: +"Optional scope override — \"strategic\", \"social\", or \"operational\". Empty = engine classifies from keywords."

Changed6 schema fields changedv4.5.5May 7, 2026

addedInput schema / properties / context / description
```
Added value: +"Background context for all models."
```

addedInput schema / properties / max_rounds / description

Added value: +"Maximum rounds (default 3 for debate, 6 for dialogue)."

addedInput schema / properties / mode / description

Added value: +"\"dialogue\" (short turns) or \"debate\" (long essays)."

addedInput schema / properties / question / description
```
Added value: +"The question to reach consensus on."
```

addedInput schema / properties / save_path / description

Added value: +"Optional file path to save the full transcript."

addedInput schema / properties / scope

Added value: +{
+  "default": "",
+  "description": "Optional scope override — \"strategic\", \"social\", or\n\"operational\". When empty, the engine classifies from\nkeywords in the question and context. Strategic and social\nscopes enforce the 3-model minimum (charter consensus-thresholds)\nand allow Grok as a tiebreaker on deadlock.",
+  "type": "string"
+}

First observedv0.1.0Mar 22, 2026

TDQS

A4.7/5.0

Behavior5/5

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

Annotations only provide readOnlyHint=false and destructiveHint=false, so the description carries most of the behavioral burden. It discloses side effects (transcript writes under save_path), external model calls through configured providers, tier-dependent model availability, and special constraints for strategic/social scopes including a 3-model minimum and possible Grok tiebreaker. This goes well beyond the bare annotations.

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

Conciseness5/5

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

The description is well-structured with clear headers for use cases, exclusions, sibling contrast, and side effects. It front-loads the core purpose and every sentence contributes distinct decision-relevant information, making the length justified rather than padded.

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

Completeness5/5

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

Given the tool's complexity, the description plus full input schema and output schema provide everything an agent needs: required parameter, supported modes, scope behavior, provider configuration, side effects, and when to avoid this tool. No critical information is missing.

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

Parameters3/5

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

Schema description coverage is 100%, so the schema already documents all 7 parameters. The description adds no new parameter-level syntax or format details beyond what the schema provides, and therefore remains at the baseline 3 rather than needing to compensate.

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?

The description opens with a specific verb phrase, 'Run multi-model consensus via AI-to-AI deliberation,' and names the exact resource. It explicitly differentiates from delimit_models and delimit_security_deliberate, so an agent can distinguish this tool from its siblings without opening their schemas.

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

Usage Guidelines5/5

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

The description provides an explicit 'When to use' list with concrete decision types such as pricing, naming, copy framing, doctrine edits, and external PR diffs. It also gives a clear 'When NOT to use' exclusion for routine implementation choices and names alternatives like in-thread orchestration or subagent dispatch.

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

Deploy Server