Skip to content

Daily Usage

Once you’ve initialized lexicon (Quickstart), here are the workflows you’ll use most often.

Most days you only need two commands:

Terminal window
lexicon verify # run gates + compute score
lexicon # start a chat session (or resume the last one)

lexicon (or lexicon chat) is the primary interface. It starts an interactive session where the AI guides you through designing contracts, gates, tests, and implementation prompts.

Terminal window
lexicon auth login # one-time setup
lexicon # start or resume a chat session

Within the chat session, the AI can:

  • Create and update contracts, gates, conformance tests, behavior scenarios
  • Infer contracts from existing code
  • Check coverage gaps and API drift
  • Sync CLAUDE.md, check repo health, manage prompts
  • Generate implementation prompts from session or existing contracts

Sessions are automatically saved and can be resumed when you restart. On resume, you’ll see the session topic and where you left off.

If an action fails during a session (e.g., referencing a contract that doesn’t exist), the AI catches the error and suggests corrections — you don’t need to debug action failures manually.

After making changes, verify everything still passes:

Terminal window
lexicon verify

This runs your gates (formatting, linting, tests), computes the quality score, checks coverage, and detects API drift — all in one command.

For a deeper check including repo health:

Terminal window
lexicon verify --health

This adds health checks for manifest, gates, scoring model, CLAUDE.md, and API baseline.

Terminal window
lexicon verify --health

Start a chat session and ask the AI to scan your API and suggest a contract:

Terminal window
lexicon

Then tell it: “Scan the public API and suggest a contract for the new module.”

Start a chat session and ask for a coverage report and improvement suggestions:

Terminal window
lexicon

Then use directives like COVERAGE_REPORT, API_SCAN, or ask for general advice.