An autonomous quality engineering swarm that forges production-ready code through continuous behavioral verification and self-healing fix loops.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install forge
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install forge using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Forge?
Forge is a sophisticated quality engineering framework within the Openclaw Skills ecosystem that unifies Domain-Driven Design (DDD), Architecture Decision Records (ADR), and Test-Driven Development (TDD) with Behavior-Driven Development (BDD). It functions as a self-learning swarm of eight specialized agents that collaborate to discover project architectures, verify backend health, and execute exhaustive end-to-end testing without the use of mocks. By strictly adhering to real API interactions, Forge identifies integration bugs that traditional mocking strategies often miss.
This skill transforms the definition of done by requiring that code not only compiles but behaves exactly as specified in Gherkin scenarios. It adapts to various topologies, including microservices, monoliths, and monorepos, ensuring consistent quality across complex software ecosystems. Forge uses a tiered confidence system to apply autonomous fixes, reducing the manual burden on developers while maintaining high safety standards through a 7-gate enforcement process.
Forge Use Cases
- Automated generation of Gherkin specifications from existing code implementations.
- Continuous behavioral verification for microservices and modular monoliths without mocking.
- Self-healing bug fixes with autonomous, confidence-tiered git commits.
- Exhaustive accessibility and security audits integrated into the quality engineering cycle.
- Validation of cross-context dependencies and API contracts to prevent integration regressions.
How Forge Works
- The swarm performs project discovery to identify backend technologies, API protocols, and bounded context boundaries.
- It verifies the backend is healthy and seeds real test data via API calls, explicitly forbidding any mocking or stubbing.
- Gherkin specifications are verified or generated for the target context to establish behavioral requirements.
- Specialized agents execute the test suite, prioritizing tests predicted to fail based on historical defect data.
- Failure Analyzer identifies root causes and matches them against a memory of fix patterns with established confidence tiers.
- Bug Fixer applies code changes ranging from simple defensive patterns to first-principles logic fixes.
- Quality Gate Enforcer evaluates the fix against seven rigorous gates including security, accessibility, and contract compliance.
- Auto-Committer stages the verified changes with detailed documentation, while the Learning Optimizer updates the swarm's intelligence.
Forge Setup
Forge is designed to work out-of-the-box with auto-discovery, but can be initialized using the following command:
npx @claude-flow/cli@latest swarm init --topology hierarchical --max-agents 10
You can optionally provide a forge.config.yaml at the root of your repository to override model routing or architecture settings. To launch a full autonomous run, use the following Openclaw Skills command:
/forge --autonomous --all
Forge Data Schema & Taxonomy
Forge maintains structured intelligence across multiple namespaces to facilitate autonomous decision-making. Data is organized as follows:
| Namespace | Metadata Purpose |
|---|---|
| forge-patterns | Stores fix patterns categorized by confidence tiers (Platinum, Gold, Silver, Bronze). |
| forge-results | Captures structured test execution data, stack traces, and Gherkin mappings. |
| forge-state | Tracks coverage status, quality gate evaluations, and the last known green commit. |
| forge-specs | Manages BDD feature files and the mapping between scenarios and test functions. |
| forge-predictions | Records historical failure rates and file change impact for defect probability. |
| forge-contracts | Maintains snapshots of API schemas to detect breaking contract regressions. |
Forge Advanced Features
- Multi-model routing that delegates tasks to specific LLM tiers (Haiku, Sonnet, Opus) to optimize for cost and reasoning depth.
- Chaos testing capabilities that inject timeouts and rate-limiting to verify application resilience.
- Visual regression testing with pixel-by-pixel comparison and screenshot diff storage for UI-heavy contexts.
- Cascade re-testing logic that identifies and verifies all downstream dependencies affected by a code fix.
- Optional Agentic QE integration for enhanced vector-indexed pattern search and full SAST/DAST security scanning.
SKILL.md
Loading
Related Openclaw Skills
Fabrik-Codek
A private cognitive architecture that builds a local knowledge graph from your workflows to personalize LLM interactions.
safe-edit
A safety-first utility that creates automatic rollbacks and backups before modifying critical configuration files.
WHOOP API Fitness Tracker
A powerful integration for retrieving and analyzing personalized WHOOP fitness data, including recovery, sleep, and strain metrics via optimized Python scripts.
ERC-8004 Identity Skill
A comprehensive framework for establishing on-chain AI agent identity, reputation, and task monetization on the Avalanche network.
Quick Note
A streamlined tool for rapidly recording inspiration and managing temporary notes directly in your local environment.
Nimble Web Search
A real-time web intelligence tool for AI agents featuring 8 specialized focus modes and AI-powered result synthesis.
Featured*
