Assemble by Cohesium AI

What is Assemble by Cohesium AI?

Assemble by Cohesium AI lets you orchestrate multiple AI specialist roles with a configuration setup across coding platforms from one place. Rather than relying on a single, generic assistant, it’s designed to coordinate different roles across planning, implementation, QA, security, and other workflows.

The project is distributed as a command-line tool you run via npx cohesiumai-assemble. It generates native configuration files that “turn your existing IDE setup into a structured AI team.” To keep configuration consistent, Assemble produces a single source-of-truth file named .assemble.yaml.

Assemble targets multiple AI coding platforms by outputting tool-specific configuration artifacts for a range of supported tools. The tool is intended to let you maintain one structured configuration and then deploy it into the native formats each coding platform expects.

Key Features

• One command to generate configuration artifacts: Run npx cohesiumai-assemble to install and generate the configuration outputs.
• Single source-of-truth configuration: A wizard creates a .assemble.yaml file, reducing the need to maintain many separate configs that can drift over time.
• Native configuration output for many platforms: Assemble generates tool-specific configuration files for 21 platforms (including Cursor, Claude Code, GitHub Copilot, Gemini CLI, Windsurf, Codex, and others listed on the site).
• Structured multi-phase execution for complex tasks: For higher-stakes work, Assemble uses a spec-driven methodology in phases—SPECIFY, PLAN, TASKS, IMPLEMENT, CLOSE—with Jarvis assessing complexity automatically via /go.
• “YOLO mode” for uninterrupted runs: The site describes a mode that executes all phases without stopping for validation and returns only when delivery is complete.
• Configuration-based approach with no runtime/service: Assemble is described as having no daemon, no server, and no process—it works by generating configuration rather than running continuously.

How to Use Assemble by Cohesium AI

1. Run the CLI tool. Start with:
   • npx cohesiumai-assemble
2. Complete the wizard to create your source configuration. Assemble uses a wizard to produce a .assemble.yaml file. This file acts as the single source of truth for the “AI team” setup.
3. Review and manage .assemble.yaml. Since this is the central configuration, changes here are what should drive the generated outputs.
4. Generate platform-native configuration files. After your .assemble.yaml is set up, Assemble outputs the native config artifacts for the supported AI coding platforms (targeting 21 platforms as stated on the site).
5. Use phase-based execution when needed. For complex tasks, follow the methodology described by Assemble’s phases—SPECIFY, PLAN, TASKS, IMPLEMENT, CLOSE—and note that Jarvis is referenced as assessing complexity automatically via /go.
6. If your workflow supports it, consider “YOLO mode.” The site’s “YOLO mode” description indicates an uninterrupted run that executes all phases without stopping for validation and returns when delivery is complete.

Note: The page describes these behaviors at a feature level; it does not provide detailed step-by-step instructions for every platform’s generated configuration within the extracted content.

Use Cases

• Standardize an “AI team” across multiple coding platforms: Use .assemble.yaml as the single configuration source and generate native configs for a range of tools.
• Coordinate multi-role workflows beyond a single assistant: Set up distinct responsibilities across planning, implementation, QA, security, and other phases mentioned in the description.
• Handle complex tasks with structured phases: Apply the SPECIFY, PLAN, TASKS, IMPLEMENT, CLOSE workflow for higher-stakes or multi-step work.
• Run without intermediate validation for delivery-focused iterations: Use the described “YOLO mode” when you want the system to proceed through all phases without stopping for validation.
• Reduce configuration drift in IDE/tool setups: Instead of maintaining many separate platform configs, generate them from .assemble.yaml.

FAQ

What does Assemble generate?

Assemble generates native configuration files for supported AI coding platforms, based on a single source-of-truth configuration file called .assemble.yaml.

How do I start using Assemble by Cohesium AI?

The page states it is provided as a command-line tool that you run using npx cohesiumai-assemble.

How many platforms are supported?

The extracted content states that Assemble generates native configuration output for 21 platforms.

What are the execution phases?

For complex tasks, Assemble describes a spec-driven methodology using the phases: SPECIFY, PLAN, TASKS, IMPLEMENT, CLOSE.

What does “YOLO mode” mean?

According to the site description, “YOLO mode” executes all phases without stopping for validation and returns only when delivery is complete.

Alternatives

• Single-assistant configuration (per IDE/tool): Many AI coding platforms rely on one assistant-like configuration per tool; you would manage separate settings directly inside each platform rather than generating configs from a shared .assemble.yaml.
• Workflow orchestration via task runners or automation scripts: Instead of configuration generation across platforms, you can orchestrate planning/implementation/QA steps using custom scripts or CI-style workflows.
• Multi-agent frameworks (run-time orchestration): Category alternatives include frameworks that coordinate multiple roles during execution, rather than using configuration artifacts to adapt existing IDE/tool setups.
• IDE-specific “rules” and prompt templates: Some solutions focus on enforcing structure through prompt templates, code review steps, and role-specific instructions within one platform’s native customization options.