feat(mcp): full CLI parity — 18 tools, composite tests, validation, bug fixes (#99)

New tools:
- init_project: scaffold probe.yaml and tests/ directory
- generate_report: HTML report from JSON results file
- record: capture interactions → .probe file with configurable timeout

shutdown_device: now accepts serial (Android) in addition to udid (iOS)

run_tests: composite_devices parameter maps ALIAS=SPEC pairs to
  --composite-device flags; description now enumerates all key flags
  (--format, --parallel, --shard, --host/--token, --disable-animations,
  --video, --stream)

write_test: validates ProbeScript via parser.ParseFile before writing;
  syntax errors returned immediately without creating the file;
  description documents composite test syntax

get_report / generate_report: fixed alphabetical-sort bug — now uses
  os.Stat modification time to find the most recently written report

Docs: rewrote mcp.md with full 18-tool reference, composite test
  examples, run_tests flags table, record/report workflows, and
  troubleshooting for write_test validation errors

Tests: 5 new tests (tool count = 18, shutdown accepts udid or serial,
  write_test rejects invalid syntax, run_tests exposes composite_devices,
  init/generate_report/record present in tools/list)

Author: pbertsch

CHANGELOG.md

@@ -6,6 +6,18 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
 ## [Unreleased]
 
+## [0.9.1] - 2026-05-09
+
+### Added
+- **MCP: 3 new tools** — `init_project` (scaffold `probe.yaml` + `tests/`), `generate_report` (HTML from JSON results), `record` (record interactions → `.probe` file with configurable timeout)
+- **MCP: Android emulator shutdown** — `shutdown_device` now accepts `serial` for Android emulators in addition to `udid` for iOS simulators
+- **MCP: composite test support** — `run_tests` exposes a `composite_devices` parameter (`"A=host:port/token B=udid"`) that maps directly to `--composite-device` flags; `write_test` description documents the full `composite test` syntax
+- **MCP: `run_tests` flags documented** — tool description now enumerates key flags (`--timeout`, `--format`, `--parallel`, `--shard`, `--host/--token`, `--disable-animations`, `--video`, `--stream`)
+- **MCP: `write_test` validates before writing** — content is parsed in-process before the file is created; syntax errors are returned immediately without touching the filesystem
+
+### Fixed
+- **MCP: `get_report` used alphabetical order instead of modification time** — `get_report` and `generate_report` now find the most recently *modified* JSON file in `reports/`, not the lexicographically last one
+
 ## [0.9.0] - 2026-05-09
 
 ### Added

README.md

@@ -412,7 +412,25 @@ Studio binaries also ship as part of every GitHub release. See the dedicated [St
 
 ## MCP Server
 
-For AI agents (Claude Desktop, Cursor, any MCP-compatible client), v0.6.0 ships **`probe-mcp`** as a standalone binary alongside `probe`. It exposes 10 tools: `get_widget_tree`, `take_screenshot`, `read_test`, `write_test`, `run_script`, `run_tests`, `list_files`, `lint`, `get_report`, `generate_test`.
+**`probe-mcp`** is a standalone binary that exposes all FlutterProbe capabilities to AI agents (Claude Desktop, Cursor, any MCP-compatible client) via 18 tools:
+
+| Category | Tools |
+|---|---|
+| Device lifecycle | `list_devices`, `list_simulators`, `list_avds`, `start_device`, `shutdown_device` |
+| Authoring | `get_widget_tree`, `read_test`, `write_test`, `run_script` |
+| Execution | `run_tests`, `list_files`, `lint`, `take_screenshot` |
+| Reporting | `get_report`, `generate_report`, `generate_test` |
+| Project | `init_project`, `record` |
+
+Every CLI feature is accessible from MCP. Key capabilities an agent can use:
+
+- **Composite tests** — `write_test` supports the full `composite test` syntax; `run_tests` accepts a `composite_devices` parameter (`"A=host:port/token B=udid"`)
+- **Structured results** — pass `--format json` in `run_tests`'s `flags` to get machine-readable output, then call `get_report` or `generate_report` for HTML
+- **Parallel/shard** — `--parallel` or `--shard 1/3` via `flags`
+- **WiFi physical devices** — `--host <ip> --token <tok>` via `flags`
+- **Android + iOS shutdown** — `shutdown_device` accepts `udid` (iOS) or `serial` (Android)
+- **Pre-write validation** — `write_test` parses the content before writing; syntax errors are returned immediately without creating the file
+- **Interaction recording** — `record` captures gestures and returns a `.probe` file
 
 Configure your MCP client:
 
@@ -426,7 +444,7 @@ Configure your MCP client:
 }
 ```
 
-The legacy `probe mcp-server` subcommand still works for backwards compatibility but prints a one-time deprecation notice. Full setup guide for Claude Desktop, Cursor, and other MCP clients: [MCP Server docs](https://flutterprobe.dev/tools/mcp/).
+The legacy `probe mcp-server` subcommand still works but prints a deprecation notice. Full setup guide for Claude Desktop, Cursor, and other MCP clients: [MCP Server docs](https://flutterprobe.dev/tools/mcp/).
 
 ## Visual Regression
 

docs/wiki/Home.md

@@ -18,7 +18,7 @@ Welcome to the FlutterProbe wiki. This documentation covers architecture details
 
 ## Project Status
 
-FlutterProbe is in active development. Current version: **0.9.0**.
+FlutterProbe is in active development. Current version: **0.9.1**.
 
 ### Repository Structure