feat: launch Vibeflow n8n v0.5.0

clients/claude-code/README.md

@@ -0,0 +1,27 @@
+# Claude Code
+
+## Goal
+
+Use Vibeflow n8n as a reusable instruction layer for building n8n workflows through MCP.
+
+## Recommended setup idea
+
+- connect Claude Code to the n8n MCP server
+- add the prompt from `templates/system-prompt.md` to your skill or instructions layer
+- keep the docs folder available so the agent can reference the behavior contract
+
+## Suggested operating mode
+
+The best default is `balanced` mode:
+- few but useful questions
+- planning before build
+- explicit assumptions
+- readable delivery report
+
+## Prompt seed
+
+```text
+Follow the Vibeflow n8n skill from this repository. Gather the minimum viable requirements, produce a normalized plan, create or update the workflow through MCP, validate it, and give me a concise final report.
+```
+
+See `config-snippets.md` for example `claude mcp add --scope project` and `.mcp.json` setup.

clients/claude-code/config-snippets.md

@@ -0,0 +1,36 @@
+# Claude Code config snippets
+
+These snippets are examples. Replace placeholder URLs and names with your real n8n MCP configuration.
+
+## Add a project-scoped HTTP MCP server
+
+```bash
+claude mcp add --transport http --scope project n8n https://YOUR-N8N-MCP-ENDPOINT
+```
+
+This writes a `.mcp.json` file at the project root.
+
+## Example `.mcp.json`
+
+```json
+{
+  "mcpServers": {
+    "n8n": {
+      "type": "http",
+      "url": "https://YOUR-N8N-MCP-ENDPOINT"
+    }
+  }
+}
+```
+
+## Optional instruction layer
+
+```text
+Use the Vibeflow n8n behavior in this repository. Ask only essential workflow questions, generate a normalized plan before any n8n changes, build through the n8n MCP server, and return a concise final report.
+```
+
+## Notes
+
+- `local` scope is private to your project entry inside `~/.claude.json`.
+- `project` scope creates a versionable `.mcp.json`.
+- `user` scope makes the server available across projects.

clients/claude-code/sample-mcp.json

@@ -0,0 +1,12 @@
+{
+  "mcpServers": {
+    "n8n": {
+      "command": "npx",
+      "args": ["-y", "n8n-mcp"],
+      "env": {
+        "N8N_BASE_URL": "http://localhost:5678",
+        "N8N_API_KEY": "replace-me"
+      }
+    }
+  }
+}

clients/codex/README.md

@@ -0,0 +1,28 @@
+# Codex CLI
+
+## Goal
+
+Use Vibeflow n8n with Codex CLI as the behavior layer that guides intake, planning, build, validation, and final reporting.
+
+## Recommended setup idea
+
+- connect Codex CLI to the n8n MCP server
+- load the core prompt from `templates/system-prompt.md`
+- optionally keep `docs/skill-spec.md` and `docs/conversation-contract.md` in the working directory for extra context
+
+## Suggested working pattern
+
+1. Start Codex in a project folder that contains this repository.
+2. Ensure MCP access to n8n is configured.
+3. Ask for a workflow in natural language.
+4. Let the agent ask a few focused questions.
+5. Review the plan.
+6. Approve the build.
+
+## Helpful prompt seed
+
+```text
+Use the Vibeflow n8n skill in this repository. Interview me briefly, produce a build plan, then create the workflow in n8n through MCP and return a final handoff report.
+```
+
+See `config-snippets.md` for example `codex mcp add` and `config.toml` setup.

clients/codex/config-snippets.md

@@ -0,0 +1,31 @@
+# Codex CLI config snippets
+
+These snippets are examples. Replace placeholder URLs and names with your real n8n MCP configuration.
+
+## Option 1: add with CLI
+
+```bash
+codex mcp add n8n --url https://YOUR-N8N-MCP-ENDPOINT
+codex mcp list
+```
+
+## Option 2: add in config file
+
+Project-scoped or user-scoped configuration can live in `~/.codex/config.toml` or `.codex/config.toml`.
+
+```toml
+[mcp_servers.n8n]
+url = "https://YOUR-N8N-MCP-ENDPOINT"
+```
+
+## Optional instruction in AGENTS.md
+
+```text
+Always use the n8n MCP server when the request is about creating, updating, validating, or testing n8n workflows. Before building anything, produce a normalized plan that matches schemas/plan.schema.json.
+```
+
+## Suggested first prompt
+
+```text
+Use the Vibeflow n8n skill in this repository. Interview me briefly, produce a normalized plan, then build the workflow in n8n through MCP and finish with a concise handoff report.
+```

clients/codex/sample-config.toml

@@ -0,0 +1,10 @@
+# Example Codex CLI MCP configuration
+# Adjust command, args, and environment to match your local n8n MCP setup.
+
+[mcp_servers.n8n]
+command = "npx"
+args = ["-y", "n8n-mcp"]
+
+[mcp_servers.n8n.env]
+N8N_BASE_URL = "http://localhost:5678"
+N8N_API_KEY = "replace-me"

clients/openclaude/README.md

@@ -0,0 +1,11 @@
+# OpenClaude / Community Forks
+
+This target is community / best-effort.
+
+If your fork supports MCP, adapt one of the sample configurations from the primary clients and point it at your n8n MCP server.
+
+Recommended approach:
+- start from the Claude Code or OpenCode examples
+- verify your client supports local MCP servers
+- confirm environment variable names for the n8n MCP process
+- test a simple read-only operation before creating workflows

clients/openclaude/config-snippets.md

@@ -0,0 +1,23 @@
+# OpenClaude and community forks: config notes
+
+This target is best-effort.
+
+Use these guidelines only if your fork supports:
+- MCP connections
+- custom instruction or skill layers
+- conversational follow-up questions
+
+## Minimal strategy
+
+1. Point your client to the n8n MCP endpoint.
+2. Load `templates/system-prompt.md`.
+3. Keep `docs/conversation-contract.md` and `schemas/plan.schema.json` available.
+4. Test with one simple recipe before using real workflows.
+
+## Compatibility contract
+
+The client should be able to:
+- use an MCP server by name
+- ask follow-up questions
+- emit a structured plan before building
+- complete a final handoff report

clients/opencode/README.md

@@ -0,0 +1,21 @@
+# OpenCode
+
+## Goal
+
+Use this repository as the instruction pack for building n8n workflows conversationally through MCP.
+
+## Suggested setup
+
+- configure the n8n MCP server in OpenCode
+- provide `templates/system-prompt.md` as the primary behavior file
+- optionally pin `docs/skill-spec.md` for extra grounding
+
+## Good defaults
+
+For most users:
+- mode: balanced
+- assumptions: explicit
+- plan required before build: yes
+- final report: concise but complete
+
+See `config-snippets.md` for example `opencode mcp add` and `opencode.jsonc` setup.

clients/opencode/config-snippets.md

@@ -0,0 +1,31 @@
+# OpenCode config snippets
+
+These snippets are examples. Replace placeholder URLs and names with your real n8n MCP configuration.
+
+## Guided setup
+
+```bash
+opencode mcp add
+opencode mcp list
+```
+
+## Example `opencode.jsonc`
+
+```jsonc
+{
+  "$schema": "https://opencode.ai/config.json",
+  "mcp": {
+    "n8n": {
+      "type": "remote",
+      "url": "https://YOUR-N8N-MCP-ENDPOINT",
+      "enabled": true
+    }
+  }
+}
+```
+
+## Suggested agent instruction
+
+```text
+When the task is about n8n workflows, use the n8n MCP server and follow the Vibeflow n8n repository contract. Produce a normalized plan first, then build, validate, and report.
+```

clients/opencode/sample-opencode.jsonc

@@ -0,0 +1,13 @@
+{
+  // Example OpenCode configuration
+  "mcp": {
+    "n8n": {
+      "type": "local",
+      "command": ["npx", "-y", "n8n-mcp"],
+      "env": {
+        "N8N_BASE_URL": "http://localhost:5678",
+        "N8N_API_KEY": "replace-me"
+      }
+    }
+  }
+}