2.6k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill opentangl- _meta.json622 B
- SKILL.md8.4 KB
Overview
This skill sets up and runs OpenTangl — an autonomous development engine for JavaScript/TypeScript projects. It scans your codebase, creates a product vision, proposes and executes tasks, verifies builds/tests, opens PRs, and merges changes in a closed loop. It works with new or existing JS/TS projects and supports multi-project environments.
How this skill works
OpenTangl reads a human-authored product vision and scans specified source directories to detect project type, package manager, and build/test scripts. It uses an LLM to propose prioritized tasks, implements code changes, runs verification commands, and generates PRs via the GitHub CLI; an LLM reviews and merges clean PRs. The loop updates the vision doc with status and repeats per configured cycles and feature/maintenance ratios.
When to use it
- Kick off a new JS/TS project scaffolded for React, Next, Express, or Serverless.
- Automate feature development and routine maintenance for an existing codebase.
- Run continuous autonomous iterations to keep a product aligned with a vision.
- Coordinate related frontend/backend repositories under a single product environment.
- Prototype rapid changes without manual PR creation, review, and merging.
Best practices
- Verify prerequisites first: Node ≥18, git with remote, authenticated gh CLI, and an LLM API key; install any missing tools before proceeding.
- Create a clear, human-authored product vision (Origin & Direction) and keep initial priorities focused (3–5 items).
- Accurately list scan_dirs and skip_patterns in projects.yaml to avoid scanning irrelevant files.
- Start with --cycles 1 to validate behavior and review the first run before increasing automation.
- Use feature-ratio to balance feature work vs maintenance (e.g., 0.8 = 80% features).
- Keep an eye on escalated PRs and build failures; adjust vision or verification commands when needed.
Example use cases
- Scaffold a new TypeScript React app with Vite and let OpenTangl propose initial features and tests.
- Point OpenTangl at an existing monorepo to detect projects, generate projects.yaml, and autonomously deliver small features across services.
- Run periodic autopilot cycles to fix lint/typecheck issues and backfill tests according to the product vision.
- Use multi-project environments to coordinate API and frontend work with a single vision document.
- Prototype a new UI flow: OpenTangl creates branches, implements changes, runs build/test, and opens PRs for review and merge.
FAQ
You need Node.js ≥18, git configured with a remote, the GitHub CLI authenticated (gh auth status), and an LLM API key (OPENAI_API_KEY or ANTHROPIC_API_KEY). If anything is missing, install it and re-run the checks.
How do I control how much feature work vs maintenance is done?
Use the --feature-ratio flag (0–1). For example --feature-ratio 0.8 makes 80% of proposed work feature-focused and 20% maintenance/testing.
Can I run OpenTangl on multiple projects at once?
Yes. Define multiple projects in projects.yaml and run autopilot with --projects id1,id2. For multi-project products, add an environment field to group projects under one vision.