feat(mcp): add MCP server with in-process command execution

[pcfreak30]

Summary

• MCP adapter exposes CLI subcommands as MCP tools via in-process execution (root.Run with buffer capture, no subprocess fork)
• FSM-based wizard sessions for website onboarding and setup flows
• Step input types with JSON schemas for structured wizard navigation
• MCP resources: account status, website DNS/validation, session state
• MCP prompt templates for website-onboarding and setup workflows
• Factory pattern: pkg/cli provides concrete wizard implementations
• Output formatter propagates root writer for in-process MCP capture

---

Description

This PR adds a Model Context Protocol (MCP) server to pinner-cli, enabling AI agents to discover and invoke CLI commands as MCP tools, access live data via pinner:// resources, and drive interactive wizard workflows through structured prompts.

Key additions:

MCP Server & Tool Adapter (pkg/internal/mcp/adapter.go): Exposes all non-hidden CLI subcommands as MCP tools over stdio. Tool invocations execute the command tree in-process (no subprocess fork), capturing stdout/stderr into buffers and returning results to the agent. Extended flag-type support beyond the upstream adapter to include Float, Duration, and StringSlice flags.

MCP Resources (pkg/internal/mcp/resources.go): Registers four pinner:// resources that agents can read at runtime:

• pinner://account/status — auth state, quota, and masked config summary
• pinner://websites/{domain}/dns-requirements — DNS records the user must configure
• pinner://websites/{id}/validation-status — live website validation result
• pinner://wizard/{session_id}/state — current wizard FSM state and next-step schema

MCP Prompts (pkg/internal/mcp/prompts.go): Two prompt templates that generate step-by-step instructions for agents:

• website-onboarding — guides the agent through the website creation wizard (auth check, content source, target type, domain, DNS mode, create, DNS setup, validate)
• setup — guides the agent through initial setup (auth, config, shell completion, tutorial)

Wizard Session Engine (pkg/internal/mcp/session.go, wizard.go): A TTL-bounded, in-memory session store with FSM-driven step progression. Each step validates JSON input, mutates wizard state, and returns a StepResponse containing the next step name and its expected input schema. Sessions support retry on handler errors (FSM stays in the same state) and abort transitions from any step. Two wizard types are implemented: websites creation and initial setup.

CLI Integration: The mcp subcommand is wired into the root command. Output formatters now respect the root command's writer, allowing in-process execution to capture output into buffers when invoked by the MCP adapter.

[github-actions]

Code Coverage Report

Total Coverage: 54.4%

Generated from commit: 737d426
Repository: LumeWeb/pinner-cli

[kody-ai]

Found critical issues please review the requested changes

• Development logger with debug-level enabled logs unredacted stderr buffer on every MCP tool call, potentially exposing sensitive data.

[kody-ai]

Kody Review Complete

Great news! 🎉
No issues were found that match your current review configurations.

Keep up the excellent work! 🚀

Kody Guide: Usage and Configuration

Interacting with Kody

• Request a Review: Ask Kody to review your PR manually by adding a comment with the @kody start-review command at the root of your PR.

• Validate Business Logic: Ask Kody to validate your code against business rules by adding a comment with the @kody -v business-logic command.

• Provide Feedback: Help Kody learn and improve by reacting to its comments with a 👍 for helpful suggestions or a 👎 if improvements are needed.

Current Kody Configuration

Review Options

The following review options are enabled or disabled:

Options	Enabled
Bug	✅
Performance	✅
Security	✅
Business Logic	✅

Access your configuration settings here.

​