Control Changelogs & Roadmaps
Directly from Your AI IDE
The official MCP server for NanoLog. Draft release notes from Git commits, update roadmap milestones, inspect live user feedback, and tweak widget settings without ever leaving Cursor, Claude Code, or Antigravity.
Watch the Complete Autonomous Loop
The video shows how an engineering team uses Claude Code with NanoLog MCP to eliminate the gap between customer feedback and product education.
User Submits In-App Feedback
Customer asks for bulk CSV upload directly via the live widget.
AI Agent Triages Reports
Developer tells Claude: "Process recent feedback". Claude pulls the top 10 reports and identifies bulk upload already exists.
Builds In-App Spotlight Tour
Claude calls create_spotlight targeting the DOM button to guide confused users.
Publishes Changelog with Tour Link
Claude drafts and publishes the update with #tour:bulk-upload, closing the loop with zero manual documentation.
One-Minute Setup in Any AI Editor
Select your preferred AI coding assistant below and copy the ready-to-run configuration.
{
"mcpServers": {
"nanolog": {
"command": "npx",
"args": ["-y", "@ahm-labs/nanolog-mcp"],
"env": {
"NANOLOG_API_URL": "https://nanolog.dev",
"NANOLOG_API_KEY": "YOUR_NANOLOG_API_KEY",
"NANOLOG_APP_ID": "YOUR_DEFAULT_APP_ID"
}
}
}
}Supercharge Your Shipping Workflow
Spend zero time writing release documentation by letting AI agents handle publishing as part of your CI and coding process.
Git-to-Changelog in One Prompt
Ask your agent: "Check my commits from today and publish a feature update to NanoLog." The agent structures notes, categorizes changes, and publishes instantly.
Autonomous Roadmap Sync
When closing PRs or completing tasks, your AI moves roadmap items from In Progress to Released without manual dashboard clicks.
Built-in Safety: Zero Deletes
Engineered with safety by design. The MCP server deliberately excludes delete tools, ensuring AI agents cannot hallucinate or wipe your release history.
14 Granular Tools for Full Control
Every aspect of your projects, posts, roadmaps, spotlights, customer feedback, and configurations exposed via typed MCP schemas.
Lists all accessible client organizations and App IDs under your multi-tenant account.
Retrieves published updates and drafts with unique IDs, categories, tags, and pin states.
Turns git diffs or feature branches into categorized, markdown-formatted product release notes.
Updates title, content, tags, pin priority, or toggles draft/published status on existing posts.
Uploads screenshots or visual assets to the Edge CDN and returns Markdown-ready image URLs.
Views roadmap items across Planned, In Progress, and Released with upvote/downvote metrics.
Logs planned feature milestones or customer bug trackers straight into your product roadmap.
Transitions status (e.g. Planned → In Progress → Released) and updates milestone descriptions.
Lists interactive DOM product tours and beacon spotlights configured for your application.
Generates multi-step in-app product tours targeting CSS selectors and links them via #tour:slug.
Updates spotlight tour steps, target DOM selectors, beacon styles, and activation states.
Inspects real-time customer bug reports, annotated screenshots, feature suggestions, and NPS scores.
Checks widget tab labels, domain security allowlists, and customer voting flags.
Configures allowed origins, localhost dev access, public pages, and widget feature toggles.
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard introduced by Anthropic and adopted across modern AI coding environments—including Claude Code, OpenAI Codex CLI, Google Antigravity, Cursor, and Windsurf. Instead of copying and pasting data between your terminal, browser, and issue tracker, MCP gives your AI coding agent secure, typed tool primitives to interact directly with external platforms.
With @ahm-labs/nanolog-mcp installed, your AI agent closes the entire product operations loop: it reads live customer bug reports (with client-side PII redaction), writes a reproduction test and fix in your codebase, generates an interactive DOM Spotlight tour for the new UI, and drafts your release notes automatically.
How to Use MCP with Claude Code
Bind NanoLog in one command (claude mcp add nanolog) to let Claude Code triage customer feedback and draft release notes from your terminal.
How to Use MCP with OpenAI Codex CLI
Configure ~/.codex/config.toml so OpenAI Codex agents can automatically sync roadmap statuses and publish changelogs after merging PRs.
How to Use MCP with Google Antigravity
Pair Antigravity's autonomous agent skills with NanoLog MCP to inspect DOM selectors, build step-by-step feature spotlights, and verify fixes with E2E tests.
How to Use MCP with Cursor & Windsurf
Drop .cursor/mcp.json into your workspace so Cursor Agent and Windsurf Cascade can turn Git commits into rich customer announcements.
MCP & AI Agent Integration FAQ
Does the AI agent publish release notes live to customers without review?
By default, NanoLog encourages a "Draft First" workflow (isPublished: false). Your AI agent drafts the release notes, attaches screenshots, and links the interactive tour in your dashboard so a human can review and publish with one click—or you can instruct the agent to publish immediately.
Can an AI agent accidentally delete my changelog history or roadmap items?
No. @ahm-labs/nanolog-mcp is architected with strict safety boundaries: it exposes Create, Read, and Update primitives only. Destructive delete endpoints are intentionally excluded from the MCP tool specification.
Which AI coding agents support the NanoLog MCP Server?
Any MCP-compliant client works out of the box via standard stdio transport (npx -y @ahm-labs/nanolog-mcp), including Claude Code, Claude Desktop, OpenAI Codex CLI, Google Antigravity, Cursor, Windsurf, and Zed.
Ready to automate your changelogs with AI?
Get started with NanoLog today. Free tier available with full MCP server access, zero style bleed widgets, and multi-tenant project management.