feat(oc_mentor): OpenClassrooms RPA + MCP server for Claude Code

.devcontainer/devcontainer.json

@@ -0,0 +1,20 @@
+{
+    "image": "mcr.microsoft.com/devcontainers/base:noble",
+    "name": "",
+    "features": {
+        "ghcr.io/tomgrv/devcontainer-features/githooks:5": {},
+        "ghcr.io/tomgrv/devcontainer-features/gitutils:5": {},
+        "ghcr.io/tomgrv/devcontainer-features/gitversion:7": {}
+    },
+    "remoteEnv": {},
+    "postCreateCommand": [
+        "mkdir -p /home/vscode/.ssh && chmod 700 /home/vscode/.ssh && ssh-keyscan github.com >>/home/vscode/.ssh/known_hosts",
+        "sudo chown -Rf vscode:vscode ${containerWorkspaceFolder:-.}/* ${containerWorkspaceFolder:-.}/.*",
+        "sudo find ${containerWorkspaceFolder:-.} -mindepth 1 -type d -exec chmod 755 {} +"
+    ],
+    "postStartCommand": [
+        "git config --global --add safe.directory ${containerWorkspaceFolder:-.}",
+        "test -z \"$CODESPACES\" && git config --global gpg.program gpg2"
+    ],
+    "customizations": {}
+}

.gitattributes

@@ -0,0 +1,8 @@
+* text=auto eol=lf
+*.{cmd,[cC][mM][dD]} text eol=crlf
+*.{bat,[bB][aA][tT]} text eol=crlf
+*.css linguist-vendored
+*.scss linguist-vendored
+*.js linguist-vendored
+*.json merge=json
+CHANGELOG.md export-ignore

.github/instructions/general.instructions.md

@@ -0,0 +1,21 @@
+---
+applyTo: '**'
+---
+
+<!-- @format -->
+
+## Minimal Changes Discipline
+
+Make the smallest possible change to address the specific request. Do not modify files unrelated to the task.
+
+- **Package files** (`package.json`, `package-lock.json`): only modify if explicitly required for the task.
+- **Build artifacts and symlinks**: do not commit unless they are the direct target of the request.
+- **Repository setup** (symlinks, environment bootstrapping): use temporarily for development/testing only; do not commit.
+- **Infrastructure changes**: avoid unless specifically requested.
+
+When working on any task:
+
+1. Identify the **exact files** that need to change.
+2. Change **only** those files.
+3. Use temporary local setup for testing; revert any unrelated side effects before committing.
+4. Focus on the **specific feature or fix** requested, not general improvements.

.github/skills/feature-common-utils/SKILL.md

@@ -0,0 +1,38 @@
+---
+name: feature-common-utils
+description: Shared shell utilities for JSON validation, normalization, and reusable zz_* helper scripts.
+---
+
+<!-- @format -->
+
+# common-utils
+
+## Description
+
+Use this feature as a shared utility layer for shell-based automation across features (`jq`, `dos2unix`, and `zz_*` helper scripts).
+
+## Commands
+
+- `validate-json [options] <json>` - Validate JSON against a schema.
+- `normalize-json [options] <json>` - Normalize JSON order/format using schema rules.
+- `zz_dist [options]` - Copy `zz_*` helpers to a target directory.
+- `zz_args` - Parse command-line arguments in shell scripts.
+- `zz_log` - Emit structured/colorized shell logs.
+- `zz_json` - Read or manipulate JSON from shell scripts.
+
+## Use For
+
+- JSON validation/normalization workflows (`validate-json`, `normalize-json`).
+- Shell scripting with standardized logging, argument parsing, and prompts (`zz_log`, `zz_args`, `zz_ask`).
+- Distributing shared helper scripts into project folders (`zz_dist`).
+
+## Do Not Use For
+
+- Feature-specific business logic.
+- Git workflow automation (use `gitutils` or `githooks`).
+
+## Agent Guidance
+
+- Reuse existing `zz_*` scripts before adding new helpers.
+- Prefer `normalize-json`/`validate-json` in lint pipelines for schema-safe edits.
+- Keep automation generic and composable for cross-feature reuse.

.github/skills/feature-gitversion/SKILL.md

@@ -0,0 +1,36 @@
+---
+name: feature-gitversion
+description: Semantic versioning utilities that derive versions from git history and update tags/changelog.
+---
+
+<!-- @format -->
+
+# gitversion
+
+## Description
+
+Use this feature when an agent must derive application version numbers from Git history in a consistent way.
+
+## Commands
+
+- `gv` - Compute version metadata from Git history.
+- `bump-version` - Update version fields/files from computed version.
+- `bump-tag` - Create/update version tag from release version.
+- `bump-changelog` - Generate or update changelog for release context.
+
+## Use For
+
+- Calculating semantic versions from branch/commit state.
+- Providing version metadata for build/release automation.
+- Aligning local version behavior with CI versioning.
+
+## Do Not Use For
+
+- Manual tag/edit workflows without GitVersion involvement.
+- Hook or lint orchestration (use `githooks`).
+
+## Agent Guidance
+
+- Use the installed GitVersion tooling as the source of truth for computed versions.
+- Keep versioning logic centralized; avoid duplicating derivation rules in scripts.
+- Pair with release helpers only when version flow is explicitly required.

.gitignore

@@ -21,3 +21,4 @@ Thumbs.db
 .idea/
 *.swp
 *.swo
+./.gitattributes

.gitversion

@@ -0,0 +1,3 @@
+major-version-bump-message: "^(build|ci|docs|feat|fix|perf|refactor|revert|style|test)(\\([\\w\\s-]*\\))?(!:|:.*\\n\\n((.+\\n)+\\n)?BREAKING CHANGE:\\s.+)"
+minor-version-bump-message: "^(feat)(\\([\\w\\s-]*\\))?:"
+patch-version-bump-message: "^(build|ci|docs|fix|perf|refactor|revert|style|test)(\\([\\w\\s-]*\\))?:"

oc_mentor/.env.example

@@ -0,0 +1,9 @@
+# OpenClassrooms credentials
+OC_EMAIL=your@email.com
+OC_PASSWORD=yourpassword
+
+# Optional: override where browser auth state is persisted (default: ~/.oc_mentor/auth-state.json)
+# AUTH_STATE_PATH=/custom/path/auth-state.json
+
+# Optional: set to "false" to run browser in headed mode for debugging
+# PLAYWRIGHT_HEADLESS=false

oc_mentor/.gitignore

@@ -0,0 +1,5 @@
+node_modules/
+dist/
+.env
+*.auth-state.json
+~/.oc_mentor/

oc_mentor/README.md

@@ -0,0 +1,125 @@
+# oc_mentor
+
+OpenClassrooms mentor RPA connector + MCP server.  
+Works as both a **Claude Code tool** (stdio) and a **Claude.ai integration** (HTTP).
+
+Exposes 8 tools backed by Playwright browser automation against the OC mentor portal:
+
+| Tool | Description |
+|---|---|
+| `oc_login` | Authenticate (called automatically on first use) |
+| `oc_status` | Auth state, student count, next session |
+| `oc_list_sessions` | List upcoming/past sessions |
+| `oc_get_session` | Session detail + notes |
+| `oc_update_session_notes` | Write session notes |
+| `oc_list_projects` | List student projects (pending/evaluated) |
+| `oc_get_project` | Project brief, deliverables, competencies |
+| `oc_submit_evaluation` | Submit pass/fail evaluation with feedback |
+
+---
+
+## Setup
+
+### 1. Install
+
+```bash
+npm install
+npx playwright install chromium
+```
+
+### 2. Configure credentials
+
+```bash
+cp .env.example .env
+# Edit .env — set OC_EMAIL and OC_PASSWORD
+```
+
+### 3. Build
+
+```bash
+npm run build
+```
+
+---
+
+## Use with Claude Code (stdio)
+
+Add to `~/.claude/settings.json`:
+
+```json
+{
+  "mcpServers": {
+    "oc_mentor": {
+      "command": "node",
+      "args": ["/absolute/path/to/oc_mentor/dist/index.js"],
+      "env": {
+        "OC_EMAIL": "your@email.com",
+        "OC_PASSWORD": "yourpassword"
+      }
+    }
+  }
+}
+```
+
+Restart Claude Code — the `oc_*` tools will appear in the tool list.
+
+---
+
+## Use as a Claude.ai integration (HTTP)
+
+Claude.ai supports remote MCP servers via its **Integrations** settings.
+
+### Start the HTTP server
+
+```bash
+# Set a secret token to protect the endpoint
+MCP_AUTH_TOKEN=mysecrettoken npm run start:http
+# → [oc_mentor] HTTP MCP server → http://localhost:3000/mcp
+```
+
+For production, run behind a reverse proxy (nginx, Caddy) with TLS.  
+Set `PORT=<n>` to change the port.
+
+### Add to Claude.ai
+
+1. Open **claude.ai → Settings → Integrations**
+2. Click **Add custom integration**
+3. Enter the server URL: `https://your-server.com/mcp` (or `http://localhost:3000/mcp` for local)
+4. Set authentication: **Bearer token** → paste the value of `MCP_AUTH_TOKEN`
+5. Save — the `oc_*` tools will appear in Claude.ai conversations
+
+### Test the endpoint
+
+```bash
+curl -X POST http://localhost:3000/mcp \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer mysecrettoken" \
+  -d '{"jsonrpc":"2.0","method":"initialize","id":1,"params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1"}}}'
+```
+
+---
+
+## Verify stdio mode
+
+```bash
+echo '{"jsonrpc":"2.0","method":"tools/list","id":1}' | node dist/index.js
+# → JSON with all 8 tools
+```
+
+---
+
+## Debug (headed browser)
+
+```bash
+PLAYWRIGHT_HEADLESS=false npm run start:http
+```
+
+---
+
+## Session persistence
+
+Auth state is saved to `~/.oc_mentor/auth-state.json` after first login and reused for up to 8 hours.
+
+## Selector maintenance
+
+OC's frontend may update its DOM — if tools stop returning data, run in headed mode and use DevTools to update selectors in `src/connector/`.

oc_mentor/package.json

@@ -0,0 +1,28 @@
+{
+  "name": "oc_mentor",
+  "version": "1.0.0",
+  "description": "OpenClassrooms mentor RPA connector + MCP server for Claude Code and Claude.ai",
+  "type": "module",
+  "main": "dist/index.js",
+  "scripts": {
+    "build": "tsc",
+    "start": "node dist/index.js",
+    "start:http": "MCP_TRANSPORT=http node dist/index.js",
+    "dev": "tsx src/index.ts",
+    "dev:http": "MCP_TRANSPORT=http tsx src/index.ts"
+  },
+  "engines": {
+    "node": ">=18"
+  },
+  "dependencies": {
+    "@modelcontextprotocol/sdk": "^1.12.0",
+    "dotenv": "^16.4.5",
+    "playwright": "^1.49.0",
+    "zod": "^3.24.0"
+  },
+  "devDependencies": {
+    "@types/node": "^22.0.0",
+    "tsx": "^4.19.0",
+    "typescript": "^5.7.0"
+  }
+}