* feat: expose framework primitives via API gateway and MCP
Add registry, store, and broker as both HTTP routes and MCP tools
so AI agents and HTTP clients can inspect and operate the framework.
API gateway (/micro/* namespace):
GET /micro/registry List registered services
GET /micro/registry/{name} Describe a service
GET /micro/store List store keys
GET /micro/store/{key} Read a record
POST /micro/store/{key} Write a record
POST /micro/broker/{topic} Publish a message
MCP gateway (micro_* tool prefix):
micro_registry_list List services
micro_registry_get Describe a service
micro_store_list List keys
micro_store_read Read a record
micro_store_write Write a record
micro_broker_publish Publish a message
Framework tools use a Handler field on the MCP Tool struct for
direct dispatch (no RPC). Service tools continue to use RPC.
Rate limiters and circuit breakers are applied to framework
tools the same as service tools.
* fix: make framework internals opt-in on API and MCP gateways
Framework primitives (registry, broker, store) are now only
exposed when explicitly enabled:
API gateway: micro api --internal
MCP gateway: Options{Internal: true}
Off by default — user services are always exposed, framework
internals require the flag. Banner output only shows framework
routes when enabled.
* fix: always expose framework internals, gate by auth in production
Revert the --internal flag approach. Framework primitives (registry,
broker, store) are now always exposed:
- micro api: /micro/* routes always available (dev tool)
- MCP gateway: micro_* tools always registered. When Auth is
configured (production), they require micro:admin scope.
Without Auth (dev), they're open — same as all other tools.
This follows the existing pattern: micro run/api = dev (open),
micro server = production (auth + scopes). Framework internals
follow the same security model as user services.
Remove the Internal option from MCP Options. Remove --internal
flag from micro api.
Note: scope persistence depends on the store backend. The default
in-memory store does not survive restarts. Use MICRO_STORE=file
for persistent scopes in production.
* fix: correct DefaultStore comment — it's file-backed, not memory
* fix(server): don't recreate deleted admin user on restart
When the default admin account is deleted via the dashboard, set
a marker key (auth/.admin-deleted) in the store. On startup, skip
admin creation if the marker exists. This prevents the default
admin/micro credentials from reappearing after restart when the
user has intentionally removed them.
* fix: improve agent playground first-run UX and fix doc 404s
Agent playground:
- Add setup hint in empty state explaining how to get started
(click Settings, enter API key, type a prompt)
- Hide hint automatically when API key is already configured
- Add all 7 providers to dropdown (was only OpenAI + Anthropic)
- Include CLI fallback suggestion (micro chat)
Docs:
- Fix .md links to .html across all doc pages — Jekyll serves
.html files, not .md. Fixes 404s including the micro run guide.
---------
Co-authored-by: Claude <noreply@anthropic.com>
* docs: add AI provider integration guide and Supported AI Providers section
Add a step-by-step guide for AI infrastructure companies to implement
ai.Model and contribute a provider to go-micro. Covers the full
lifecycle: skeleton, tool call handling, tests, registration, and PR
checklist.
Add a "Supported AI Providers" section to the project README that lists
current providers (Anthropic, OpenAI) in a table and links to the
integration guide with a call-to-action for new providers and sponsors.
Streamline the "Adding a New Provider" section in ai/README.md to point
to the new guide instead of duplicating a full code listing.
* fix: remove nonexistent Discord link from README
* fix(website): set content container width to 800px on desktop
Move the 800px max-width from .markdown-body up to .content so
the entire content pane (not just the inner body) is sized
correctly. The container now fills up to 800px beside the sidebar.
* feat(ai): wire Atlas Cloud into server and auto-detection
Import atlascloud provider in the micro server so it is available
when running micro run / micro server. Add atlascloud to
AutoDetectProvider so --ai_base_url with an atlascloud domain
selects the right provider automatically.
* feat(ai): add Google Gemini provider
Add ai/gemini implementing ai.Model for Google's Gemini API. Uses
the native generateContent endpoint with system_instruction,
contents/parts, and functionDeclarations — not an OpenAI shim.
Default model gemini-2.5-flash, auth via x-goog-api-key header.
Wire into micro server imports and AutoDetectProvider (matches
googleapis.com and google in base URL).
Update README.md and ai/README.md with provider listing.
* feat(ai): add Groq, Mistral, and Together AI providers
Add three new OpenAI-compatible providers:
- ai/groq: ultra-fast inference, default model llama-3.3-70b-versatile
- ai/mistral: Mistral AI, default model mistral-large-latest
- ai/together: Together AI, default model Llama-3.3-70B-Instruct-Turbo
All three are wired into the micro server imports and
AutoDetectProvider. README and ai/README updated with the full
provider table.
* feat(ai): add ai/tools helper and 'micro chat' interactive agent
Extract the registry-discovery + RPC-execution loop from the web
agent playground into a reusable ai/tools package:
- tools.New(reg) creates a Set bound to a registry
- Set.Discover() walks the registry and returns []ai.Tool with
LLM-safe (underscored) names, remembering the mapping back to
the original dotted form
- Set.Handler(client) returns an ai.ToolHandler that resolves
the safe name and issues the RPC
Add cmd/micro/chat — an interactive 'micro chat' REPL that uses
ai/tools to let users talk to their services through any
registered AI provider. Supports --prompt for single-shot use,
auto-detects the provider from --base_url, and falls back to the
provider's conventional env var (ANTHROPIC_API_KEY, etc).
Update README with the new command and the programmatic example.
* feat(examples): add gRPC interop example
Add examples/grpc-interop showing that any standard gRPC client can
call a go-micro service — no go-micro SDK required on the client
side. Includes:
- proto/greeter.proto with generated Go, gRPC, and micro stubs
- server/ using go-micro gRPC transport
- client/ using stock google.golang.org/grpc (no go-micro imports)
- README with Python example and explanation of how routing works
Addresses the confusion from issue #2818 where users didn't know
that go-micro gRPC services are callable by any gRPC client.
* fix: strip /api prefix from MCP routes
Change /api/mcp/tools and /api/mcp/call to /mcp/tools and
/mcp/call. MCP is a first-class feature, not a sub-path of the
API proxy. Update server routes, playground template, scopes
template, run.go output, README, CLI README, and all docs.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Redesign the /agent playground with improved UX:
- Chat-focused layout with full-height message area and sticky input
- Collapsible tool call cards showing name, input, result, and timing
- Thinking indicator while waiting for agent response
- Settings panel collapsed by default (auto-opens if no API key)
- Empty state with available tools preview
- Clear chat button
- Better visual hierarchy with distinct message styles
Co-authored-by: Claude <noreply@anthropic.com>
* Initial plan
* Add MCP Playground page to web UI with tool discovery and calling
- Add playground.html template with chat-style agent prompt interface
- Add /playground route handler in server
- Add /api/mcp/tools endpoint to list available MCP tools
- Add /api/mcp/call endpoint to invoke MCP tools via RPC
- Add Playground link to sidebar navigation
- Playground auto-discovers services from registry and renders them as
callable tools with input forms and activity logging
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Add playground.html template and fix .gitignore micro pattern
Fix .gitignore pattern 'micro' -> '/micro' to only ignore root-level
binary, not paths containing 'micro' as a component.
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Address code review: use crypto/rand for trace IDs, fix var redecl, remove dup comment
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Rename Playground to Agent, fix styling, add LLM-powered prompt
- Rename /playground to /agent, move Agent link to top of sidebar menu
- Fix template styling: use existing form/input/button CSS from styles.css
instead of inline styles and form-plain class
- Add /api/agent/settings GET/POST endpoints for model API key, model
name, and base URL configuration (stored in server store)
- Add /api/agent/prompt POST endpoint that sends user prompt to
OpenAI-compatible LLM API with tool definitions from registry,
executes any tool calls via RPC, and returns results with a
follow-up LLM summary
- Show available tools in a table using existing table styles
- Prompt section is placed above settings for primary workflow
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Address code review: handle unmarshal errors, extract system prompt, improve param descriptions
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Add Anthropic model support to the agent
Support both OpenAI and Anthropic APIs in the agent prompt handler:
- Add provider selector (OpenAI/Anthropic) to settings UI and backend
- Auto-detect provider from base URL when not explicitly set
- Anthropic: use /v1/messages endpoint, x-api-key header, input_schema
format for tools, content blocks for responses, tool_use/tool_result
message format for follow-ups
- OpenAI: unchanged /v1/chat/completions with Bearer auth
- Default models: gpt-4o (OpenAI), claude-sonnet-4-20250514 (Anthropic)
- Provider-specific defaults for base URLs
Co-authored-by: asim <17530+asim@users.noreply.github.com>
* Remove duplicate Anthropic follow-up message construction
Co-authored-by: asim <17530+asim@users.noreply.github.com>
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: asim <17530+asim@users.noreply.github.com>
Asim Aslam