Automatic Stack Detection

Relevant source files

The following files were used as context for generating this wiki page:

• cli/src/config/detector.ts
• cli/src/config/writer.ts
• cli/src/config/loader.ts
• cli/src/config/types.ts
• morty.sh
• README.md

Automatic Stack Detection is a core initialization feature in Morty that scans a target project's root directory to identify its programming language, framework, and standard development commands (test, lint, and build). This system ensures that when Morty is initialized in a "brownfield" or existing project, the AI agents are immediately aware of the project's technical context and can execute relevant CLI tools without manual configuration.

Sources: cli/src/config/detector.ts:16-24, morty.sh:162-171, README.md:46-51

Detection Architecture

The detection logic is triggered during the initialization phase (via morty --init). It operates by looking for specific "sentinel" files associated with major ecosystems.

Core Detection Logic

The system evaluates the project in a specific order, generally prioritizing package managers and configuration files:

1. Node.js/JavaScript/TypeScript: Triggered by package.json.
2. Python: Triggered by pyproject.toml, requirements.txt, or setup.py.
3. Go: Triggered by go.mod.
4. Rust: Triggered by Cargo.toml.

Sources: cli/src/config/detector.ts:26-55, morty.sh:176-248

Detection Flow Diagram

The following diagram illustrates the decision tree used to identify the project stack and extract relevant metadata.

Ecosystem-Specific Detection

Node.js Ecosystem

For Node.js projects, the detector parses package.json to extract the project name and identify frameworks by checking dependencies and devDependencies. It prioritizes meta-frameworks (e.g., Next.js, Nuxt, Remix) over base libraries like React or Vue. It also checks for the existence of bun.lockb to determine if bun should be used for testing instead of npm.

Sources: cli/src/config/detector.ts:58-106, morty.sh:176-227

Python, Go, and Rust

For non-Node ecosystems, the detector sets standard industry commands:

Ecosystem	Sentinel Files	Default Test Command	Default Lint Command
Python	pyproject.toml, requirements.txt, setup.py	pytest	ruff check .
Go	go.mod	go test ./...	golangci-lint run
Rust	Cargo.toml	cargo test	cargo clippy

Sources: cli/src/config/detector.ts:108-135, morty.sh:229-248

Data Persistence and Usage

Once detection is complete, the results are encapsulated in a DetectedProject object and used to populate the .morty/config.yaml file. This configuration is then used by Morty's execution loop to provide context to the AI and run automated checks.

Configuration Structure

The detected data is mapped to the following schema:

Field	Description	Source
name	Project identifier	package.json name or directory name
language	Primary language	Detection of .ts, .py, .go, .rs files
framework	Web or app framework	Dependency analysis
testCmd	Command to run tests	Script detection or ecosystem defaults
lintCmd	Command to run linting	Script detection or ecosystem defaults
buildCmd	Command to build project	Script detection or ecosystem defaults

Sources: cli/src/config/types.ts:6-22, cli/src/config/writer.ts:16-44, cli/src/config/detector.ts:4-13

Initialization Sequence

The following sequence diagram shows how the initConfig function coordinates with the detectProject utility to set up the environment.

Summary

Automatic Stack Detection is the gateway for Morty's autonomous operations in existing repositories. By accurately identifying the environment through sentinel file analysis, Morty reduces the friction of manual setup, allowing AI agents to understand the project's language, framework constraints, and preferred tooling immediately upon initialization.

Sources: cli/README.md:38-44, README.md:46-51, morty.sh:248-261