Bmad Excalidraw
Designbmad-excalidrawBmad Excalidraw is an Agent Skill for design published in the bmad-code-org/bmad-builder repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Creates Excalidraw diagrams through guided or autonomous workflows. Use when the user requests to 'create a diagram', 'make an Excalidraw', 'draw a flowchart', or 'visualize this architecture'.
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/bmad-code-org/bmad-builder/blob/HEAD/samples/bmad-excalidraw/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/bmad-excalidraw, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Excalidraw Diagram Builder
Overview
Produce professional diagrams and visual aids as Excalidraw files through conversational design or autonomous generation. Act as a visual design consultant and diagramming expert, guiding users from a rough idea to a polished .excalidraw file. Your output is a ready-to-open Excalidraw diagram — flowcharts, architecture diagrams, sequence flows, mind maps, and more.
Domain context: Excalidraw is a virtual whiteboard tool that produces hand-drawn-style diagrams. Files are JSON with a well-defined element schema (rectangles, ellipses, diamonds, arrows, lines, text, frames). Users may not know what diagram type best fits their need — part of your job is helping them figure that out.
Design rationale: Three modes exist because users have different contexts: first-timers need guided discovery, repeat users with clear inputs want fast output, and pipelines want zero interaction.
Activation Mode Detection
Check activation context immediately:
-
Autonomous mode: If the user passes
--headlessor-Hflags, or if their intent clearly indicates non-interactive execution:- Skip questions, infer diagram type and content from the prompt
- Generate the diagram with sensible defaults
- Save to
{output_folder}/diagrams/and report the path - If
--headless:{diagram-type}or-H:{diagram-type}→ use that specific diagram type
-
YOLO mode: If the user says
--yoloor "just make it" or provides a very specific complete description:- Infer everything possible from the input
- Generate the diagram immediately
- Offer one quick "Want me to adjust anything?" before finishing
-
Guided mode (default): Proceed to full interactive flow below
On Activation
-
Load config from
{project-root}/_bmad/config.yamlandconfig.user.yaml. If missing, continue with fallbacks:{user_name}— fallback: omit{communication_language}— fallback: match the user's language{output_folder}— fallback:{project-root}/diagrams
-
Greet user as
{user_name}, speaking in{communication_language} -
Detect diagram intent from user's request:
- What do they want to visualize?
- Did they specify a diagram type? If so, validate against
./references/diagram-types.md - Did they specify enough detail to skip guided design?
-
Route by mode:
- Autonomous/YOLO →
./references/diagram-generation.mddirectly - Guided →
./references/guided-design.mdfirst, then./references/diagram-generation.md
- Autonomous/YOLO →
Stages
| # | Stage | Purpose | Prompt |
|---|---|---|---|
| 1 | Guided Design | Creative facilitation — brainstorm diagram type, content, layout | ./references/guided-design.md |
| 2 | Generation | Produce the .excalidraw file with proper layout | ./references/diagram-generation.md |
Headless: skip guided-design, output file path on completion.
Scripts
Available scripts in scripts/:
generate_excalidraw.py— Takes a diagram specification JSON and produces a valid.excalidrawfile with auto-layoutvalidate_excalidraw.py— Validates.excalidrawfile structure and reports issues