Testing bramble

Skill wevial/bramble/.agents/skills/testing-bramble

Test bramble CLI features and spec output. Use when verifying CLI flags, spec format converters, or TUI behavior.From its SKILL.md

Install

npx -y skills add wevial/bramble --skill testing-bramble

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.2 KB, as published. Nobody here has run it

Testing Bramble

Prerequisites

bun must be installed (curl -fsSL https://bun.sh/install | bash, then source ~/.bash_profile)
Run bun install in the repo root
No API keys needed for fake-agent mode (the default)

Running Tests

Unit Tests

bun run test           # all tests
bun run test -- <file> # single file, e.g. src/docs/format.test.ts

Typecheck

bun run typecheck

Known Pre-Existing Test Failures

The 6 UI test files (ConversationPane, FlowSidebar, SetupScreen, SpecPane, StatusStrip, markdown) may fail with a .scm file extension error from @opentui/core. These are pre-existing and unrelated to feature changes.

Testing CLI Flags

CLI flags can be tested directly:

bun src/index.tsx --help                    # verify flag appears in help
bun src/index.tsx --format pdf 2>&1         # verify invalid values error

Testing Format Converters

The spec converters (src/docs/format.ts) can be tested via:

Unit tests in src/docs/format.test.ts
Integration scripts that import and call the converters with realistic spec content

A realistic test spec is the "Authentication Spec" that the fake agents produce (visible in the canned responses in src/index.tsx).

Testing the TUI

The TUI uses alternate-screen mode and requires a TTY. To test interactively:

bun run dev -- "design an auth system"           # fake agents
bun run dev -- --format json "design an auth system"  # test with format flag

After a session completes, check output files in .bramble/<session-name>/:

spec.md (or spec.xml, spec.json, spec.html depending on --format)
debate.md, interview.md, transcript.jsonl

Key Architecture Notes

Internal spec representation is always Markdown (agents use find/replace patches)
Format conversion happens at the persistence layer (writeSpec call in App.tsx)
sessionPaths() in src/sessions/list.ts controls file extensions
CLI flag parsing is in src/index.tsx (manual argv loop, not a flag library)