❓ Frequently Asked Questions — Early CLI

This page covers CLI-specific questions when running early generate-pr. For shared topics like API keys, token scopes, and permissions, see the Fundamentals FAQ →.

---

🌐 Which languages are supported?

• ✅ JavaScript
• ✅ TypeScript
• ✅ Python (pytest)

The CLI auto-detects the project language from file extensions and project markers (e.g., package.json for JS/TS, pyproject.toml or requirements.txt for Python).

---

🔐 What permissions do tokens need?

---

TOKEN — for Git access and committing files

Used to access commit or PR metadata, push commits, and post comments.

See Token Requirements →

---

🔑 How do I get my API key?

See API Key →

---

🧱 Which CI/CD systems are supported?

Any system with Node.js 20+:

• Jenkins
• CircleCI
• GitLab
• Bitbucket
• Self-hosted agents

Python projects additionally require pip install py-agent in the CI environment.

---

🧪 Which test frameworks are supported?

JavaScript / TypeScript

• ✅ Jest (required today)
• 🟡 Vitest (coming soon)
• ⚪ Mocha (planned)

See Framework Support →

Python

• ✅ pytest (auto-detected)

The test framework is automatically set to pytest for Python projects. No configuration is needed.

---

📈 What coverage threshold is used?

Currently fixed at 0%. Only functions with zero coverage will be targeted.

---

⚙️ How do I handle memory issues?

JavaScript / TypeScript

If you hit a heap error:

export MAX_CONCURRENCY=2

Or:

export NODE_OPTIONS="--max-old-space-size=5120"

See Memory Tuning →

Python

Python test generation does not require NODE_OPTIONS memory tuning. If you experience performance issues, reduce concurrency:

export MAX_CONCURRENCY=1

---

🧩 Can I override the coverage or test commands?

JavaScript / TypeScript

Yes. Use these environment variables:

Variable	Purpose
COVERAGE_COMMAND	Run coverage before/after
TEST_COMMAND	Run test validation

Flags like --coverageDirectory and --json must stay intact. See Configuration →

Python

Custom --test-command and --coverage-command are not yet supported for Python projects. The agent analyzes source code directly to identify testable functions.

---

🧾 How does auto-commit work?

JavaScript / TypeScript

When AUTO_COMMIT=true:

• Tests are committed to the branch
• Only passing, valid tests are committed

To disable:

export AUTO_COMMIT=false

Python

When AUTO_COMMIT=true:

• Generated .early. test files are committed and pushed to the branch
• The commit message includes [skip ci] by default (configurable via AUTO_COMMIT_SKIP_CI)

To disable:

export AUTO_COMMIT=false

---

🧹 What if my repo enforces lint/format rules?

JavaScript / TypeScript

Use:

export LINT_COMMAND="npx --no eslint $early_filename --fix"
export PRETTIER_COMMAND="npx --no prettier $early_filename --write"

These run before test validation and commit. You can customize the command, just leave the placeholders, as-is.

See Linting & Formatting →

Python

Lint and formatting hooks are not yet supported for Python projects. Generated test files follow standard pytest conventions.

---

📁 Where are test files placed?

JavaScript / TypeScript

You control this via --test-structure:

• siblingFolder → next to source
• rootFolder → inside root tests/ folder

Python

Test files are placed in .early.test/ sibling folders next to the source file:

src/
  utils/
    helpers.py
    helpers.early.test/
      test_my_function_early.py
      test_another_function_early.py

Each testable function gets its own test file.

---

🧪 Can I run this locally?

JavaScript / TypeScript

Yes. Example:

export API_KEY=...
export TOKEN=...
export HEAD_REF=my-branch
export AUTO_COMMIT=false

early generate-pr

Make sure your machine is configured to access your SCM.

See CLI Configuration →

Python

Yes. Install the CLI and py-agent, then run:

npm install -g @earlyai/cli
pip install py-agent

export API_KEY=...
export TOKEN=...
export HEAD_REF=my-branch
export AUTO_COMMIT=false
export TEST_FRAMEWORK=pytest

early generate-pr

---

🧰 What if no tests are generated?

Possible reasons:

• All changed code is already covered
• Nothing testable in this PR
• Coverage failed to collect (JS/TS) or no testable functions found (Python)

---

⚠️ Why did commit fail?

Most likely: the branch changed mid-run and your push was rejected.

Fix:

• Re-run the CLI

---

🧩 How does this differ from the GitHub Action?

Option	Usage
GitHub Action	PR workflows in GitHub
Early CLI	Any CI/CD system

Both produce identical output.

---

💬 Still need help?

• 📬 Email: support@startearly.ai
• Slack (for invited teams)