# Open-Workflow-Library: A Catalog and Toolset for n8n Automation Workflows > Open-Workflow-Library is a Python toolset and workflow collection for n8n automation that catalogs, audits, and statically validates around 1,700 source workflow files plus approximately 420 generated templates. It is a developer-grade open-source project aimed at engineers who want to search, study, or build on a large body of n8n workflows without needing n8n running locally. Workflow execution, LLM-backed generation, and multi-framework support are documented as goals but are not yet implemented. **oxbshw/Open-Workflow-Library** — Open workflow intelligence for discovering, validating, repairing, and generating automation workflows across frameworks — starting with n8n. - Repository: https://github.com/oxbshw/Open-Workflow-Library - Stars: 554 · Forks: 180 - Language: Python - License: MIT - Published: 2026-09-10 · Updated: 2026-09-10 · Language: en - Canonical page: https://hysenlabs.com/projects/oxbshw-open-workflow-library ## What Open-Workflow-Library Is and Who It Is For Automation engineers who use n8n often search for working workflow examples before writing their own. The workflows available in public repositories vary widely in quality: some are complete and correct, others reference credentials that no longer exist, embed secrets, or have structural issues that prevent import. Open-Workflow-Library addresses this by providing an audited collection with a toolset that scans for secrets, scores structural quality, and catalogs the workflows by integration and trigger type. The project description calls it open workflow intelligence for collecting, indexing, validating, repairing, and generating automation workflows across frameworks. As of the last push on 2026-05-24, the n8n collection and its associated Python tools are the implemented core. The multi-framework expansion to Dify, LangGraph, Make, and Zapier is documented in the architecture but not yet built. The intended users are n8n developers and automation engineers who want to search a large workflow collection without running n8n locally, study workflow patterns, or use the audit tools to check their own workflow files for secrets and structural problems. ## The Workflow Collection and Its Honest Status The repository contains approximately 1,700 n8n workflow source files in the workflows/ directory and an additional approximately 420 generated templates validated with the validate_generated_pack.py tool. The README includes a candid status table that distinguishes what is implemented from what is planned or out of scope. Implemented features include the audit and secret scan (which redacts matches and reported zero findings on its current run), the unified catalog with deterministic quality scoring, schema validation (using jsonschema when available, with a lightweight fallback), duplicate analysis (exact and title-similarity detection, surfacing candidates but not auto-deduplicating), and repair proposals (emitted as human-reviewable proposals, never auto-applied). The README is explicit about what is not implemented: no workflow has been imported into n8n and executed by this repository. The prompt-to-IR pipeline is a deterministic keyword-rule MVP, not an LLM-backed system. Multi-framework export to Dify, LangGraph, Make, and Zapier is documented but not built. Autonomous self-improvement (promoting learning events into the curated wiki without human review) is explicitly not in scope. Nothing in the human review queue is ever auto-approved. ## Running the Tools All tools are standard-library Python 3.12 or higher. Nothing requires global installation and no external packages are needed. n8n does not need to be running on the local machine. To audit the workflow collection, scan for secrets, and produce the catalog: ```bash python tools/audit_workflows.py ``` This produces a structured JSON report at reports/workflow-audit.json, a human-readable summary at reports/workflow-audit.md, and the indexed catalog at catalog/workflows.index.json. The tool is read-only: it never modifies, moves, or deletes workflow files, and it never prints real secret values. Detected matches against secret-like patterns are replaced with the string REDACTED before writing to disk. To build the unified catalog with quality scores: ```bash python tools/build_unified_catalog.py ``` To search the catalog by keyword: ```bash python tools/search_workflows.py --query "lead qualification" --top 5 ``` The experimental prompt-to-n8n pipeline accepts a natural-language description and attempts to generate a structural workflow file using keyword rules (not an LLM). The tool is prompt_to_n8n.py in the tools/ directory. The README links to docs/demo.md for a full walkthrough of all tools. ## What Is Not Implemented and Why That Matters The status table in the README is unusually direct about gaps. The most important gap for a practitioner is behavioral execution: no workflow in this repository has been run inside n8n. The quality score in the catalog is a deterministic structural score, not a correctness score. A workflow that passes structural validation may still fail to import into n8n or produce wrong outputs at runtime. The prompt-to-workflow pipeline is currently a keyword-rule MVP. The README describes it as deterministic: it uses a conservative node whitelist and safe placeholders. Passing a complex natural-language prompt and expecting a ready-to-run workflow is not supported; the output requires manual review and completion. Multi-framework support is the project's stated long-term goal. The Universal Workflow IR (schemas/workflow-ir.schema.json) is designed to be framework-agnostic, and the architecture documentation explicitly accommodates importers and exporters for Dify, LangGraph, Make, and Zapier. However, only n8n has practical tooling support in the current repository. A team working with any other framework gets the schemas and the documentation vision but no working importer or exporter. Repair proposals are emitted as structured JSON for human review. The system never applies them automatically. This is a deliberate design choice documented in the README, but it means the repair tooling produces reports rather than fixes. ## Comparison with n8n's Official Template Library n8n (n8n.io) maintains an official template library at n8n.io/workflows. These templates are curated, tagged with integrations, and accessible directly from the n8n UI. They have been reviewed and are importable with a single click. The practical difference is coverage and audit depth. n8n's official library is curated by the n8n team and contributors, so it represents a vetted subset of possible workflows. Open-Workflow-Library collects a larger body of community-sourced and generated workflows (over 2,100 combined) and runs a systematic audit across all of them to detect secrets, catalog integrations, and score structural quality. The breadth comes at the cost of execution-tested quality: none of the workflows in this repository have been run. For a team building n8n automations, the two resources are complementary rather than competing. n8n's official templates are the first stop for common patterns. Open-Workflow-Library is useful when the team needs a broader set of examples to study, wants to run the audit tools on their own workflow files, or wants to contribute to building a framework-agnostic workflow knowledge base. ## Repository Structure, Licence, and Maintenance The MIT licence covers the code and workflow files in the repository. The last push was on 2026-05-24 and there are no GitHub releases. The repository includes a CONTRIBUTING.md and a SECURITY.md. The top-level directory includes schemas/ (four JSON schemas), wiki/ (LLM-usable pattern and integration documentation), docs/ (architecture, supported frameworks, demo walkthrough, security model), catalog/ (the generated index), reports/ (audit output), tools/ (the five Python tools), and workflows/ (the workflow files). The audit tool's security model is documented in docs/security.md: it never prints real secret values and replaces matches with REDACTED. The project's architecture documentation describes a six-layer target: library, catalog, Universal Workflow IR, validator and repair engine, prompt-to-workflow generator, and a human-reviewed self-improvement loop. Layers one through four are the current implemented state. Layers five and six are the roadmap. Promotion of learning events into the curated wiki and rules requires human review at every step; the README states that autonomous self-improvement is not in scope. ## Conclusion Open-Workflow-Library is worth adopting for engineers who want a searchable, audited collection of n8n workflow templates and the Python tools to inspect, catalog, and structurally validate them. It is not a substitute for n8n itself, and it does not execute or behaviorally test any workflow. Before using it, verify that you have Python 3.12 or higher and that you understand the gap between structural validation and behavioral correctness: a workflow that passes schema validation may still fail when imported into n8n. The last push to the repository was on 2026-05-24. ## FAQ ### Does Open-Workflow-Library require n8n to be installed locally to run the tools? No. The README explicitly states that the tools are standard-library Python 3.12+ and nothing requires n8n on the local machine. The tools audit and catalog workflow files without executing them. Executing the workflows in n8n requires a separate n8n installation. ### Can I import workflows from Open-Workflow-Library into n8n and run them? The workflow files are in n8n's native format and can be imported, but the README states explicitly that no workflow in this repository has been imported into n8n and executed. The quality scores are structural, not correctness scores. A workflow that passes structural validation may still fail at runtime. ### What is the Universal Workflow IR and which frameworks does it currently support? The Universal Workflow IR is a framework-agnostic JSON schema defined in schemas/workflow-ir.schema.json. It is designed to represent workflows from any automation framework. As of the last push on 2026-05-24, only n8n has practical tooling support; importers and exporters for Dify, LangGraph, Make, and Zapier are described in the architecture documentation but are not yet built. ## Sources - [Issues](https://github.com/oxbshw/Open-Workflow-Library/issues) - [License: MIT](https://github.com/oxbshw/Open-Workflow-Library/blob/main/LICENSE) - [oxbshw/Open-Workflow-Library on GitHub](https://github.com/oxbshw/Open-Workflow-Library) - [README](https://github.com/oxbshw/Open-Workflow-Library/blob/main/README.md) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/oxbshw-open-workflow-library