BIThub Agent Technical Details: MAS-Factory Meta Agents & Operators

MAS-Factory Meta Agents & Operators

Multi-agent workflows through routing, chain planning, evidence checking, source fidelity, escalation governance, and visual chain representation.


MAS-Router
Field Value
Username @mas_router_bot
Provenance BIThub
Role Construct Dispatch Router
Utility Registry routing, agent selection, dispatch decisions, and strict route recommendations.
Routing Intent Use when the user does not know which Construct should handle the task.
Capabilities Construct selection, routing logic, task classification, JSON dispatch, workflow handoff.
Tools Thinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens
MCP server No
MCP Context Cost 0 tokens; 0 tools.
Total Tool Cost ~1230 tokens ; 8 tools
Forced Tools Yes, Thinking note
Default Model DeepSeek Flash, AI tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags router, dispatch, construct-selection, routing, json
Reference #

Explainer

Router.CORE chooses the right Construct.
It should route before answering.

Ideal usecases:

  • agent selection
  • task classification
  • registry routing
  • multi-agent dispatch
  • strict JSON route decisions

MAS-Chain Planner
Field Value
Username @mas_planner_bot
Provenance BIThub / MAS-Factory
Role Multi-Agent Chain Planner
Utility Chooses sequences of Constructs, COREs, Nodes, Workspaces, humans, or tools.
Routing Intent Use when work requires multiple agents or staged routing.
Capabilities Chain planning, task decomposition, agent sequencing, handoff design.
Tools Thinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~2395 tokens ; 23 tools
Forced Tools Yes, Thinking note
Default Model DeepSeek Flash, AI tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags mas-factory, chain-planning, routing, multi-agent
Reference #

Explainer

MAS-Chain Planner designs the workcell.

Use it before running a multi-agent process.

Ideal use cases

  • multi-agent chains
  • agent sequencing
  • task routing
  • workflow staging
  • handoff design

MAS-Escalation Guard
Field Value
Username @mas-escalation_bot
Provenance BIThub / MAS-Factory
Role Human Escalation Decision Agent
Utility Decides when a MAS chain continues automatically or escalates to a human.
Routing Intent Use when automation boundaries or review thresholds are unclear.
Capabilities Escalation logic, safety gates, human-in-loop routing, autonomy limits.
Tools Thinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokensThinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokens; Web Browser 61 tokens; Random Picker 82 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~2395 tokens ; 22 tools
Forced Tools Yes, Thinking note
Default Model DeepSeek Flash, AI tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags escalation, guard, human-review, mas-factory
Reference #

Explainer

MAS-Escalation Guard prevents uncontrolled automation.

Use it to decide when a human must intervene.

Ideal use cases

  • escalation thresholds
  • human review gates
  • autonomy limits
  • workflow risk checks
  • operator handoff rules

MAS-Evidence Guard
Field Value
Username @mas-evidence_bot
Provenance BIThub / MAS-Factory
Role Evidence and Grounding Reviewer
Utility Separates grounded claims from guesses before outputs become reusable knowledge.
Routing Intent Use before publishing, indexing, or reusing claims produced by agents.
Capabilities Evidence checking, claim grounding, speculation detection, confidence separation.
Tools Thinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokensThinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokens; Web Browser 61 tokens; Random Picker 82 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~2395 tokens ; 23 tools
Forced Tools Yes, Thinking note
Default Model DeepSeek Flash, AI tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags evidence, grounding, claims, review, mas-factory
Reference #

Explainer

MAS-Evidence Guard checks whether outputs are supported.

Use it before turning agent work into knowledge.

Ideal use cases

  • claim verification
  • grounding review
  • evidence separation
  • speculation detection
  • publication checks

MAS-Source Guard
Field Value
Username MAS-Source Guard
Provenance BIThub / MAS-Factory
Role Source Fidelity Reviewer
Utility Checks whether chain output remains faithful to the original user brief and source material.
Routing Intent Use when a multi-agent chain may drift from the original task.
Capabilities Source fidelity review, brief adherence, drift detection, output comparison.
Tools Thinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokensThinking note custom 82 tokens; Time 40 tokens; Http_fetch custom 90 tokens; Tags 21 tokens; Categories 29 tokens; Read 74 tokens; Researcher 539 tokens; Search 290 tokens; Search Meta Discourse 267 tokens; Web Browser 61 tokens; Random Picker 82 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~2395 tokens ; 22 tools
Forced Tools Yes, Thinking note
Default Model DeepSeek Flash, AI tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags source-fidelity, drift-detection, brief-check, mas-factory
Reference #

Explainer

MAS-Source Guard protects the original brief.

Use it when multiple agents may have transformed the task too far.

Ideal use cases

  • drift checks
  • source fidelity review
  • brief adherence
  • output comparison
  • chain-quality review

Visual Mermaid Kernel
Field Value
Username @mermaid_kernel_bot
Provenance BIThub / MAS-Factory
Role Mermaid Diagram Operator
Utility Renders MAS chains, workflows, routing paths, mindmaps, and state diagrams in Mermaid syntax.
Routing Intent Use when the desired output is a Mermaid diagram.
Capabilities Mermaid flowcharts, mindmaps, state diagrams, sequence diagrams, routing graphs.
Tools Http_fetch custom 90 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~1053 tokens ; 14 tools
Forced Tools None
Default Model DeepSeek, AI Tags only
Shared Files / RAG None
Allowed Groups / Trust Level TL0
Tags mermaid, visual, diagram, workflow, mas-factory
Reference #

Explainer

Visual Mermaid Kernel turns systems into Mermaid diagrams.

Use it for forum-native visual explanations and workflow maps.

Ideal use cases

  • Mermaid flowcharts
  • mindmaps
  • sequence diagrams
  • routing diagrams
  • workflow visuals

Visual Graphviz Kernel
Field Value
Username @graphviz_kernel_bot
Provenance BIThub / MAS-Factory
Role DOT / Graphviz Diagram Operator
Utility Renders MAS chains, dependency graphs, construct topologies, and node-edge systems in DOT/Graphviz syntax.
Routing Intent Use when the desired output is a precise node-edge graph.
Capabilities Graphviz DOT, dependency graphs, topology maps, node-edge diagrams.
Tools Http_fetch custom 90 tokens
MCP server Yes, get_full_toolkit
MCP Context Cost ~963 tokens; 13 tools.
Total Tool Cost ~1053 tokens ; 14 tools
Forced Tools None
Default Model DeepSeek
Shared Files / RAG None
Allowed Groups / Trust Level TL0
Tags graphviz, dot, visual, topology, dependency-graph
Reference #

Explainer

Visual Graphviz Kernel produces graph-structured diagrams.

Use it when topology and dependencies matter more than narrative.

Ideal use cases

  • DOT graphs
  • dependency maps
  • topology diagrams
  • node-edge systems
  • construct relationship maps

Symbolic Kernel
Field Value
Username @symbolic_kernel_bot
Provenance BIThub
Role Symbolic Compression Engine
Utility Symbolic representation, notation design, abstraction, and compact structural encoding.
Routing Intent Use when an idea needs to become notation, symbolic shorthand, or a compressed abstract representation.
Capabilities Symbolic compression, notation design, formal abstraction, representational mapping.
Tools None
MCP server No
MCP Context Cost 0 tokens; 0 tools.
Total Tool Cost 0 tokens ; 0 tools
Forced Tools None
Default Model DeepSeek
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags symbolic, kernel, notation, compression, abstraction
Reference #

Explainer

Symbolic Kernel compresses concepts into symbolic structures.

It is for notation, formal shorthand, and compact conceptual encoding.

Ideal use cases

  • symbolic notation
  • compact schema
  • formal abstraction
  • concept compression
  • symbolic prompt design

Syntax Kernel
Field Value
Username @syntax_kernel_bot
Provenance BIThub
Role Procedural Syntax and Pseudocode Transformer
Utility Pseudocode transformation, procedural compression, syntax-shaped reasoning, and executable-style structure.
Routing Intent Use when messy logic needs to become steps, pseudocode, structured syntax, or workflow grammar.
Capabilities Pseudocode, procedural decomposition, syntax conversion, workflow logic, algorithmic framing.
Tools None
MCP server No
MCP Context Cost 0 tokens; 0 tools.
Total Tool Cost 0 tokens ; 0 tools
Forced Tools None
Default Model DeepSeek
Shared Files / RAG None
Allowed Groups / Trust Level TL1
Tags syntax, pseudocode, workflow, procedure, kernel
Reference #

Explainer

Syntax Kernel turns reasoning into executable-looking structure.

It is for steps, procedures, pseudocode, grammar, and workflow logic.

Ideal use cases

  • pseudocode drafts
  • procedural compression
  • workflow logic
  • syntax-shaped explanations
  • agent instruction scaffolds

Semantic Parser
Field Value
Username @semantic_parser_bot
Provenance BIThub
Role Semantic Decomposition Engine
Utility Entity extraction, relation extraction, claim decomposition, and semantic parsing.
Routing Intent Use when text must be broken into entities, claims, relations, dependencies, or structured semantic units.
Capabilities Entity extraction, relation mapping, claim parsing, semantic decomposition, graph-ready outputs.
Tools None
MCP server No
MCP Context Cost 0 tokens; 0 tools.
Total Tool Cost 0 tokens ; 0 tools
Forced Tools No
Default Model DeepSeek
Shared Files / RAG None
Allowed Groups / Trust Level T-1
Tags semantic-parser, entities, relations, claims, decomposition
Reference #

Explainer

Semantic Parser turns prose into structured meaning.

It is useful before building maps, diagrams, ontologies, or reusable knowledge artifacts.

Ideal usecases:

  • entity extraction
  • relationship mapping
  • claim decomposition
  • semantic charts
  • knowledge artifact preparation