- Home
- Skills
- Technical Documentation
- draft architecture
draft architecture
Transform project documentation and ADRs into validated LikeC4 architecture models and identifies documentation gaps.
Free
Works with the AI Agents you already use
draft architecture
Example session with this skill installed
Draft the architecture from the docs in the /docs folder and create a LikeC4 model in /architecture.
- Read your context and instructions
- Compiled the draft architecture
I have analyzed the documentation in /docs, including the ADRs and the system-design.md file. The system appears to be a microservices-based e-commerce platform focusing on order processing and inventory management.
Architecture Draft Results
| Subsystem | Source | Code Mapping | Status |
|---|---|---|---|
| Gateway | docs/design.md#api | src/gateway | Draft |
| OrderManager | docs/adr-004.md | src/orders | Draft |
| InventoryDB | docs/design.md#data | service: postgres | Draft |
| PaymentQueue | docs/design.md#events | service: rabbitmq | Draft |
Model Statistics
- Elements: 6 (1 Person, 2 Systems, 3 Containers)
- Views: 3 (System Context, Container, Order Flow)
- Validation: Passed (LikeC4 v1.59.4)
Documentation Gaps Identified
Missing Relationship: OrderManager sends data to InventoryDB, but the database documentation does not list OrderManager as an authorized client.
- Outdated Config:
docker-compose.ymlreferences alegacy-authservice not mentioned in any design docs.
Next steps
- Review the proposed
architecture/model.c4andarchitecture/GAPS.mdfiles. - Run
npx likec4@1.59.4 start architecture/to preview the diagrams in your browser. - Install
archtracevianpm install -g archtraceto check these rules against your actual source code.
Connects securely to your tools. The creator never sees your data.
About this skill
Documentation quickly falls out of sync with the actual system design. This skill bridges that gap by transforming your READMEs, ADRs, and API specs into a structured LikeC4 architecture model. It treats your documentation as the source of truth, creating a verifiable blueprint that ensures your code aligns with your design intent.
What it does
- Extracts architecture from design docs, ADRs, and config files without reading source code.
- Generates LikeC4 models including specifications, subsystem definitions, and dynamic views.
- Identifies documentation gaps by flagging contradictions or missing connections in a
GAPS.mdfile. - Maps elements to code by associating architectural components with specific directory structures and services.
- Validates syntax using the LikeC4 CLI to ensure the generated model is error-free.
How it works
- Discovery phase where the agent scans your
docs/orarchitecture/folders and summarizes findings. - Proposal step where it lists identified subsystems and waits for your confirmation.
- Model generation to create
.c4files for elements, views, and relationships with cited sources. - Code mapping where it matches architectural elements to your project's folder names and infrastructure services.
Frameworks & tools
Built for the LikeC4 modeling language and C4 architecture patterns. It utilizes Node.js for validation and integrates with the archtrace CLI for automated drift detection.
Why this beats prompting it yourself
Writing C4 models manually is tedious and prone to syntax errors. This skill strictly enforces source-to-model traceability, ensuring every arrow and box is backed by a specific line in your documentation rather than hallucinated assumptions.
Use cases
- Onboarding new developers by generating clear, up-to-date system diagrams from existing docs.
- Audit readiness by creating a verifiable map of system dependencies and data flows.
- Drift detection by checking if the current code structure violates the documented architecture.
Known limitations
Requires Node.js 20 or newer. It focuses on documentation analysis and will not scan internal source code logic to determine architecture.
Requires Node.js 20+. Focuses on documentation analysis; does not infer architecture by reading internal source code logic.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 14 days ago
- Free with an Agensi account
Requires