DZ23 Subagents Universal MCP

Parallel specialist swarm

swarm_run

Run up to seven role-specific specialists in parallel for a project goal, then consolidate their outputs with an integrating reviewer into a structured answer.

Instructions

Run up to seven advisory specialists in parallel, then one integrating reviewer. May consume provider credits and take minutes. Reuse the project_id and mission_id of the current task (from project_init / memory_checkpoint); when omitted, results go to project "default" and a new random mission that mission_status and memory_checkpoint will not find.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`goal`	Yes	Project goal for every specialist.
`roles`	No	Specialist roles; workers cycle through this list.
`detail`	No	Controls response verbosity.	normal
`context`	No	Workspace evidence. Requires workspace; content is untrusted data.
`privacy`	No	Secrets in attached context are always masked. auto also masks valid CPF/CNPJ/card numbers, emails and phones; local_only additionally routes only to local private targets (no_local_target otherwise); allow_cloud keeps personal data.	auto
`workspace`	No	Configured workspace root for context.
`max_agents`	No	Number of workers; defaults to one per role, capped by DZ23_MAX_CONCURRENCY.
`mission_id`	No	Mission of the current task; omitted creates a new random mission.
`project_id`	No	Project of the current task; defaults to "default".	default
`lease_token`	No	Token from mission_claim. Required only while another holder has a live lease on this mission.
`output_schema`	No	JSON Schema subset (type, required, properties, items, enum) for the answer. delegate retries once and fails with response_invalid; consensus and swarm_run add structured_output or schema_error per answer.
`response_mode`	No	summary returns the full integration plus a short excerpt per worker; full returns every worker output.	summary
`idempotency_key`	No	Caller key, valid 24 h in this process. The same key and arguments return the first result without a new provider call; different arguments fail with idempotency_conflict.
`routing_strategy`	No	How workers are assigned to eligible targets. Effective strategy and observed diversity are reported.	first
`strict_diversity`	No	Fail before any provider call when the requested diversity cannot be planned.
`max_response_chars`	No	Hard cap for returned textual content.
`min_distinct_models`	No	Desired distinct models; a warning is returned when unavailable.
`avoid_reviewer_target`	No	Prefer an integrating reviewer target not used by any worker.
`min_distinct_providers`	No	Desired distinct providers; a warning is returned when unavailable.

Schema Changelog

Changes observed during successful MCP inspections.

Changed17 schema fields changedv4.1.0Sep 17, 2026

addedInput schema / properties / context

Added value: +{
+  "additionalProperties": false,
+  "description": "Workspace evidence. Requires workspace; content is untrusted data.",
+  "properties": {
+    "files": {
+      "items": {
+        "maxLength": 4096,
+        "minLength": 1,
+        "type": "string"
+      },
+      "maxItems": 50,
+      "type": "array"
+    },
+    "git_diff": {
+      "default": false,
+      "type": "boolean"
+    },
+    "search": {
+      "additionalProperties": false,
+      "properties": {
+        "max_results": {
+          "default": 50,
+          "maximum": 100,
+          "minimum": 1,
+          "type": "integer"
+        },
+        "query": {
+          "description": "Search query.",
+          "maxLength": 2000,
+          "minLength": 1,
+          "pattern": "\\S",
+          "type": "string",
+          "x-pattern-reason": "must contain non-whitespace text"
+        },
+        "regex": {
+          "default": false,
+          "type": "boolean"
+        }
+      },
+      "required": [
+        "query"
+      ],
+      "type": "object"
+    }
+  },
+  "type": "object"
+}

addedInput schema / properties / detail

Added value: +{
+  "default": "normal",
+  "description": "Controls response verbosity.",
+  "enum": [
+    "brief",
+    "normal",
+    "full"
+  ],
+  "type": "string"
+}

addedInput schema / properties / idempotency_key

Added value: +{
+  "description": "Caller key, valid 24 h in this process. The same key and arguments return the first result without a new provider call; different arguments fail with idempotency_conflict.",
+  "maxLength": 128,
+  "minLength": 8,
+  "pattern": "^[A-Za-z0-9._:-]+$",
+  "type": "string"
+}

addedInput schema / properties / lease_token

Added value: +{
+  "description": "Token from mission_claim. Required only while another holder has a live lease on this mission.",
+  "maxLength": 128,
+  "type": "string"
+}

addedInput schema / properties / max_agents / description

Added value: +"Number of workers; defaults to one per role, capped by DZ23_MAX_CONCURRENCY."

addedInput schema / properties / max_response_chars

Added value: +{
+  "description": "Hard cap for returned textual content.",
+  "maximum": 200000,
+  "minimum": 256,
+  "type": "integer"
+}

changedInput schema / properties / mission_id / description

Previous value: -"Mission identifier inside the project."New value: +"Mission of the current task; omitted creates a new random mission."

changedInput schema / properties / mission_id / pattern

Previous value: -"^[a-zA-Z0-9][a-zA-Z0-9._-]{0,119}$"New value: +"^[a-zA-Z0-9](?:[a-zA-Z0-9._-]{0,118}[a-zA-Z0-9_-])?$"

changedInput schema / properties / mission_id / x-pattern-reason

Previous value: -"must use 1-120 letters, digits, dots, underscores or hyphens and start with a letter or digit"New value: +"must use 1-120 letters, digits, dots, underscores or hyphens, start with a letter or digit and not end with a dot"

addedInput schema / properties / output_schema

Added value: +{
+  "description": "JSON Schema subset (type, required, properties, items, enum) for the answer. delegate retries once and fails with response_invalid; consensus and swarm_run add structured_output or schema_error per answer.",
+  "type": "object"
+}

addedInput schema / properties / privacy

Added value: +{
+  "default": "auto",
+  "description": "Secrets in attached context are always masked. auto also masks valid CPF/CNPJ/card numbers, emails and phones; local_only additionally routes only to local private targets (no_local_target otherwise); allow_cloud keeps personal data.",
+  "enum": [
+    "auto",
+    "local_only",
+    "allow_cloud"
+  ],
+  "type": "string"
+}

changedInput schema / properties / project_id / description

Previous value: -"Project memory identifier. Not an authorization boundary."New value: +"Project of the current task; defaults to \"default\"."

changedInput schema / properties / project_id / pattern

Previous value: -"^[a-zA-Z0-9][a-zA-Z0-9._-]{0,119}$"New value: +"^[a-zA-Z0-9](?:[a-zA-Z0-9._-]{0,118}[a-zA-Z0-9_-])?$"

changedInput schema / properties / project_id / x-pattern-reason

Previous value: -"must use 1-120 letters, digits, dots, underscores or hyphens and start with a letter or digit"New value: +"must use 1-120 letters, digits, dots, underscores or hyphens, start with a letter or digit and not end with a dot"

addedInput schema / properties / response_mode

Added value: +{
+  "default": "summary",
+  "description": "summary returns the full integration plus a short excerpt per worker; full returns every worker output.",
+  "enum": [
+    "summary",
+    "full"
+  ],
+  "type": "string"
+}

addedInput schema / properties / roles / description

Added value: +"Specialist roles; workers cycle through this list."

addedInput schema / properties / workspace

Added value: +{
+  "description": "Configured workspace root for context.",
+  "maxLength": 4096,
+  "type": "string"
+}

Changed27 schema fields changedv3.0.0Sep 15, 2026

addedInput schema / additionalProperties
```
Added value: +false
```

addedInput schema / properties / avoid_reviewer_target

Added value: +{
+  "default": false,
+  "description": "Prefer an integrating reviewer target not used by any worker.",
+  "type": "boolean"
+}

addedInput schema / properties / goal / description

Added value: +"Project goal for every specialist."

addedInput schema / properties / goal / maxLength
```
Added value: +8000
```
addedInput schema / properties / goal / minLength
```
Added value: +1
```
addedInput schema / properties / goal / pattern
```
Added value: +"\\S"
```
addedInput schema / properties / goal / x-pattern-reason
```
Added value: +"must contain non-whitespace text"
```
changedInput schema / properties / max_agents / maximum
```
Previous value: -32New value: +7
```

addedInput schema / properties / min_distinct_models

Added value: +{
+  "description": "Desired distinct models; a warning is returned when unavailable.",
+  "maximum": 7,
+  "minimum": 1,
+  "type": "integer"
+}

addedInput schema / properties / min_distinct_providers

Added value: +{
+  "description": "Desired distinct providers; a warning is returned when unavailable.",
+  "maximum": 7,
+  "minimum": 1,
+  "type": "integer"
+}

addedInput schema / properties / mission_id / description

Added value: +"Mission identifier inside the project."

addedInput schema / properties / mission_id / maxLength
```
Added value: +120
```
addedInput schema / properties / mission_id / minLength
```
Added value: +1
```

addedInput schema / properties / mission_id / pattern

Added value: +"^[a-zA-Z0-9][a-zA-Z0-9._-]{0,119}$"

addedInput schema / properties / mission_id / x-pattern-reason

Added value: +"must use 1-120 letters, digits, dots, underscores or hyphens and start with a letter or digit"

addedInput schema / properties / project_id / default
```
Added value: +"default"
```

addedInput schema / properties / project_id / description

Added value: +"Project memory identifier. Not an authorization boundary."

addedInput schema / properties / project_id / maxLength
```
Added value: +120
```
addedInput schema / properties / project_id / minLength
```
Added value: +1
```

addedInput schema / properties / project_id / pattern

Added value: +"^[a-zA-Z0-9][a-zA-Z0-9._-]{0,119}$"

addedInput schema / properties / project_id / x-pattern-reason

Added value: +"must use 1-120 letters, digits, dots, underscores or hyphens and start with a letter or digit"

addedInput schema / properties / roles / default

Added value: +[
+  "architect",
+  "backend",
+  "frontend",
+  "security",
+  "qa",
+  "devops",
+  "reviewer"
+]

addedInput schema / properties / roles / items / enum

Added value: +[
+  "architect",
+  "backend",
+  "frontend",
+  "security",
+  "qa",
+  "devops",
+  "reviewer"
+]

addedInput schema / properties / roles / maxItems
```
Added value: +7
```
addedInput schema / properties / roles / minItems
```
Added value: +1
```

addedInput schema / properties / routing_strategy

Added value: +{
+  "default": "first",
+  "description": "How workers are assigned to eligible targets. Effective strategy and observed diversity are reported.",
+  "enum": [
+    "first",
+    "round_robin",
+    "provider_diversity",
+    "model_diversity",
+    "cost_optimized",
+    "latency_optimized"
+  ],
+  "type": "string"
+}

addedInput schema / properties / strict_diversity

Added value: +{
+  "default": false,
+  "description": "Fail before any provider call when the requested diversity cannot be planned.",
+  "type": "boolean"
+}

First observedv2.2.4Sep 12, 2026

TDQS

A3.8/5.0

Behavior4/5

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

It adds valuable behavioral context beyond the annotations: 'May consume provider credits and take minutes' warns about cost/latency, and the mission/project default behavior tells the agent that omitted IDs create undiscoverable state. No contradiction with readOnlyHint=false or openWorldHint=true.

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?

Three sentences with no filler: action, cost/time warning, and the critical ID caveat are front-loaded. It avoids repeating schema details despite the large parameter surface.

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

Completeness3/5

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

For a 19-param tool with no output schema, the description covers high-level orchestration and the most important default behavior, and the schema covers the rest. However, it does not describe the return value or when to choose this over delegate/consensus, which are non-trivial gaps for a complex tool.

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 description coverage is 100%, so the baseline is 3. The description adds useful cross-tool semantics for project_id/mission_id, such as sourcing them from project_init/memory_checkpoint and the unfindable-mission consequence of omission, which the schema alone does not convey. Other parameters are appropriately left to the schema.

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 specific action and resource: 'Run up to seven advisory specialists in parallel, then one integrating reviewer.' This clearly conveys the orchestration role and distinguishes it from single-agent tools like delegate, though it does not explicitly name sibling alternatives.

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

Usage Guidelines3/5

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

The only usage guidance is the warning to reuse project_id/mission_id and the consequence of omission. It does not state when to prefer swarm_run over delegate/consensus or give exclusion criteria, so selection context is only implied by 'parallel specialists.'

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

Parallel specialist swarm

Instructions

Input Schema

Schema Changelog

TDQS

Other Tools