io.github.alexeibespalov/trident

Measured conclusion: Deep-tested. L2 outcome: tools_ok.

Create and edit collaborative architecture diagrams with any AI assistant using the Trident 2D DSL.

Add to your agent

Copy this into your MCP client config:

{
  "mcpServers": {
    "trident": {
      "url": "https://mcp.tridentchart.com/mcp",
      "transport": "streamable-http"
    }
  }
}

Measured tools (23)

• get_trident_spec read-only

  Returns the complete Trident 2D specification including grammar, syntax rules, coordinate system, containers, nodes, connections, shapes, and icon reference. Use this when you need deep understanding of the Trident DSL.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• get_generator_guide read-only

  Returns LLM-optimized instructions for generating Trident diagrams, including quick reference, critical rules (especially Y-axis!), spacing guidelines, common patterns, color conventions, and error prevention checklist. Use this before generating Trident markup (a whole diagram as text). Not needed up front in a live session that edits a document with add_node / add_connection — load it there only for domain conventions (BPMN/UML/flowchart layout, ERD, C4).

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• get_comprehensive_example read-only

  Returns a comprehensive Trident example demonstrating all syntax features including containers, nodes (traditional and bracket-style syntax), diamond/rectangle shapes, icons, connections with labels, routing modes, cards, and text annotations. Use this to see all features in action.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• validate_trident read-only

  Returns a comprehensive validation checklist for Trident diagrams covering syntax rules, semantic rules, best practices, and common mistakes to avoid. Use this after generating a diagram to verify correctness.

  Parameters:

  • diagram — string — Optional: The Trident diagram code to include context-specific validation tips

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• get_quick_start read-only

  Returns a condensed 2-minute quick-start guide with minimal working examples, core syntax reference, and key concepts. Use this for rapid learning when you need to generate simple diagrams quickly.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• open_document read-only

  Opens a live Trident document and returns its full contents as Trident markup DSL — the human-readable text format used to author diagrams. Use this to READ and UNDERSTAND the diagram: its structure, labels, connections, and layout. Do NOT rely on this to enumerate entity IDs for programmatic use — the DSL can be very large and the output may be truncated. To get a complete, structured list of all entity IDs and counts, use get_document_summary instead. Requires a valid access token.

  Parameters:

  • token (required) — string — Access token for this document. The document ID is embedded in the token.
  • agentId — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• get_document_summary read-only

  Returns a complete structured summary of a Trident document as JSON: all node IDs, container IDs, connection IDs, annotation IDs, labels, and entity counts. Use this for PROGRAMMATIC OPERATIONS — it is never truncated regardless of diagram size, so you always get every entity ID. Use this before any bulk update, delete, or enumeration task. To read and understand the diagram as a human-readable diagram, use open_document instead. Requires a valid access token.

  Parameters:

  • token (required) — string — Access token for this document. The document ID is embedded in the token.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• get_recent_changes read-only

  Returns the recent change history for a live Trident document. Each entry shows who made the change, what was changed, and when. Useful for understanding what edits have been made by AI agents or human collaborators. Requires a valid access token.

  Parameters:

  • token (required) — string — Access token for this document. The document ID is embedded in the token.
  • limit — number — Optional: Maximum number of entries to return (default 100, max 500)
  • since — number — Optional: Unix timestamp in milliseconds — only return changes after this time

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• add_node

  Adds a new node (entity) to a live Trident document. The node appears immediately for all collaborators. Requires a valid editor access token. Before adding nodes: call open_document to understand the diagram layout and pick sensible positions; call get_document_summary to get all existing entity IDs so you can avoid duplicates. IMPORTANT: if this node belongs inside a container, pass node.container on THIS call — do NOT create the node without a container and reparent it later via update_node. Orphaned nodes appear immediately to all live collaborators and create unnecessary visual churn.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • node (required) — object — Node definition. ERD field rows: when node.container is an ERD table (a container with kind:"erd"), the node becomes that table's next row — give only id, label (e.g. "🔑 id", "🔗 user_id", "email") and container, and do NOT pass x/y: the row is placed and the table resized for you.
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent no · open-world yes

• update_node

  Updates properties of an existing node in a live Trident document. Only the fields you provide are changed; others are preserved. Use get_document_summary to get all node IDs before calling this. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • nodeId (required) — string — ID of the node to update
  • changes (required) — object — Fields to update. Only provided fields are changed.
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent yes · open-world yes

• delete_node destructive

  Removes a node from a live Trident document. All connections to/from this node are also removed automatically. Use get_document_summary to get all node IDs before calling this. This action is reversible via Ctrl+Z in the Trident editor. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • nodeId (required) — string — ID of the node to delete
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive yes · idempotent yes · open-world yes

• add_container

  Adds a new container (group/swimlane) to a live Trident document. Containers visually group related nodes. Use get_document_summary to check existing container IDs before adding to avoid duplicates. Requires a valid editor access token. To build an ERD (entity-relationship diagram), add one container per table with kind:"erd" and an x/y position, then add each field with add_node (container:<table id>, no x/y) — rows are stacked and the table sized automatically — and join TABLES (never rows) with add_connection using an ERD pair type such as "one_to_many".

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • container (required) — object
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent no · open-world yes

• update_container

  Updates properties of an existing container in a live Trident document. Only provided fields are changed. Use get_document_summary to get all container IDs before calling this. Requires a valid editor access token. This is also how you RESIZE a container: set width/height (and x/y to keep it centred over its nodes) — do this whenever you add or move nodes outside a fixed-size container.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • containerId (required) — string — ID of the container to update
  • changes (required) — object
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent yes · open-world yes

• delete_container destructive

  Removes a container from a live Trident document. By default, nodes inside are detached (kept on the diagram without a container). Pass deleteNodes: true to remove the container's nodes as well. Use get_document_summary to get all container IDs before calling this. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • containerId (required) — string — ID of the container to delete
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • deleteNodes — boolean — If true, also delete all nodes inside the container (default false)
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive yes · idempotent yes · open-world yes

• add_connection

  Adds a connection (arrow/edge) between two nodes or containers in a live Trident document. Use get_document_summary to get valid source and target IDs before calling this. Requires a valid editor access token. For ERD relationships, connect the two TABLE containers (field rows are rejected as endpoints) and set type to an ERD pair such as "one_to_many".

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • connection (required) — object
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent no · open-world yes

• delete_connection destructive

  Removes a connection from a live Trident document. Provide either the connectionId, or both source and target node IDs to find and remove the matching connection. Use get_document_summary to get all connection IDs before calling this. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document. The document ID is embedded in the token.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • source — string — Source node ID (use with target if connectionId unknown)
  • target — string — Target node ID (use with source if connectionId unknown)
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • connectionId — string — ID of the connection to delete (preferred)
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive yes · idempotent yes · open-world yes

• get_document_png read-only

  Renders the current state of a live Trident document as a PNG image directly from the Yjs collaborative session — bypassing Firestore, which may be stale. Returns a base64-encoded PNG. Use this to visually verify that diagram edits look correct before or after making changes.

  Parameters:

  • token (required) — string — Access token for this document. The document ID is embedded in the token.

  Hints: read-only yes · destructive not declared · idempotent not declared · open-world yes

• add_annotation

  Adds a text annotation to a live Trident document. Two visual styles: style:stickyNote (default — post-it card with colored background, good for notes/callouts) or style:textBody (plain transparent text, good for diagram titles and section labels). Ideal for architectural notes, decision records, section labels, or TODO markers. Appears immediately for all collaborators. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • annotation (required) — object — Annotation definition
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent no · open-world yes

• update_annotation

  Updates an existing text annotation in a live Trident document. Only the fields you provide are changed. Use get_document_summary to get all annotation IDs before calling this. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • annotationId (required) — string — ID of the annotation to update
  • changes (required) — object — Fields to update. Only provided fields are changed.
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive no · idempotent yes · open-world yes

• delete_annotation destructive

  Removes a text annotation from a live Trident document. Use get_document_summary to get all annotation IDs before calling this. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • annotationId (required) — string — ID of the annotation to delete
  • userName — string — First name of the human you are assisting (ask them at session start if you do not know). Shown in the collaboration presence dot as "[agentId] working with [userName]".
  • action_explanation — string — Optional: brief explanation of what this action does and why (max 250 chars). Shown live to human collaborators in the AI cursor tooltip.

  Hints: read-only no · destructive yes · idempotent yes · open-world yes

• update_canvas_graphic

  Repositions or resizes a background image (canvas graphic) in a live Trident document. Use open_document to get the canvas graphic ID and current dimensions — it appears as: image <id> "<url>" at (x, y) width:W height:H. The position at (x, y) is the TOP-LEFT corner. USE THIS BEFORE placing many geographically-positioned nodes: if the background map is small (e.g. width < 700), scale it up so nodes have room to breathe and do not overlap. Requires a valid editor access token.

  Parameters:

  • token (required) — string — Editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot").
  • canvasGraphicId (required) — string — ID of the canvas graphic to update (from open_document output).
  • changes (required) — object — Geometry fields to update. Only provided fields are changed; all others are preserved.
  • userName — string — First name of the human you are assisting.
  • action_explanation — string — Optional: brief explanation shown live to human collaborators (max 250 chars).

  Hints: read-only no · destructive no · idempotent yes · open-world no

• make_suggestion

  Posts a visible suggestion to human collaborators watching the document WITHOUT making any changes to the diagram. Use this when you observe something worth flagging — a potential mistake, an improvement, or a question — but want the human to decide. The suggestion appears as a bold "SUGGESTION:" callout in the UI with a radar pulse animation on your presence dot for 10 seconds. Does NOT mutate the diagram. Requires a valid viewer or editor access token.

  Parameters:

  • token (required) — string — Viewer or editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • userName (required) — string — First name of the human you are assisting.
  • suggestion (required) — string — The suggestion text to show to human collaborators (max 500 chars). Be concise and actionable.
  • entityId — string — ID of the node or container this suggestion relates to. Always provide this when the suggestion concerns a specific element — it moves the AI cursor to that element so collaborators know exactly what you are referring to.

  Hints: read-only no · destructive no · idempotent no · open-world yes

• explain read-only

  Delivers an explanation payload to human collaborators watching the document, optionally anchored to a specific node or container. Use this when you want to explain what a diagram element represents, why it exists, or how it relates to other parts of the system — without suggesting a change. The explanation appears in the UI attributed to you. Does NOT mutate the diagram. Requires a valid viewer or editor access token. IMPORTANT: this tool automatically pauses (3–15 s, proportional to explanation length) before returning, so the human has time to read. Do NOT add your own artificial delays between explain calls — the pacing is built in.

  Parameters:

  • token (required) — string — Viewer or editor access token for this document.
  • agentId (required) — string — YOUR AI assistant name (e.g. "Claude.ai", "GitHub Copilot"). Shown as "[agentId] working with [userName]" in the collaboration presence dot.
  • userName (required) — string — First name of the human you are assisting.
  • explanation (required) — string — The explanation text to deliver to human collaborators (max 1000 chars).
  • entityId — string — ID of the node or container being explained. Always provide this when explaining a specific element — it moves the AI cursor to that element so collaborators can follow along.

  Hints: read-only yes · destructive no · idempotent no · open-world no

Repository

Measured by MCPYP probing. Untested fields are omitted, never estimated. Listing is always free; MCPYP never proxies calls or stores user tokens.