GoVisual

GoVisual

Runtime HTTP debugger for Go, with an MCP server so your coding agent can debug the app too.

Wrap your handler, get a local dashboard of every request. Point Claude Code or Cursor at the MCP endpoint, and the agent can list requests, replay them against your app, diff the response before and after a fix, and read the logs, SQL queries, and panic stacks that came with each capture.

Featured In

• Applied Go Weekly, May 2025
• GDG Berlin Golang Meetup, talk: "Building GoVisual: Zero-Config HTTP Debugging in Go" (Jan 2026)
• Hacker News, Show HN front page
• Golang Weekly (X), 9,100+ impressions

What's new in 2.0

• MCP server so coding agents can read, replay, and verify against your running app.
• Zero-dependency core. The main module is stdlib plus google/pprof. Storage backends (postgres, redis, sqlite, mongodb) and telemetry are separate modules you pull in only when you use them.
• Per-request capture for SQL queries, outbound HTTP, slog output, and panic stacks, all attached to the request that caused them.
• Dashboard is loopback-only by default. Replay and system-info endpoints are opt-in.
• Dark mode, live push updates, agent activity feed.

Upgrading from v0.2.1? See Upgrading below.

Install

go get github.com/doganarif/govisual/v2

The core has no database drivers, no gRPC, no OTel SDK. Add-ons install separately:

go get github.com/doganarif/govisual/mcp              # MCP server for coding agents
go get github.com/doganarif/govisual/store/postgres   # or redis, sqlite, mongodb
go get github.com/doganarif/govisual/telemetry        # OpenTelemetry export

Quick Start

package main

import (
	"net/http"

	"github.com/doganarif/govisual/v2"
)

func main() {
	mux := http.NewServeMux()
	mux.HandleFunc("/api/users", userHandler)

	handler := govisual.Wrap(mux,
		govisual.WithRequestBodyLogging(true),
		govisual.WithResponseBodyLogging(true),
	)

	http.ListenAndServe(":8080", handler)
}

Open the dashboard at http://localhost:8080/__viz.

Give your coding agent eyes

The mcp module serves captured traffic to any MCP client. Claude Code (or Cursor, or a local script) can call twelve tools against your running app: read the most recent error, pull the full crime scene for one request, replay it after you change the code, and diff the response.

package main

import (
	"net/http"

	gvmcp "github.com/doganarif/govisual/mcp"
	"github.com/doganarif/govisual/v2"
	"github.com/doganarif/govisual/v2/store"
)

func main() {
	mux := http.NewServeMux()
	// ...routes...

	st := store.WithNotify(store.NewMemory(200))
	app := govisual.Wrap(mux, govisual.WithStore(st))

	root := http.NewServeMux()
	root.Handle("/mcp", gvmcp.Handler(st, gvmcp.WithBaseURL("http://localhost:8080")))
	root.Handle("/", app)
	http.ListenAndServe(":8080", root)
}

Register with Claude Code:

claude mcp add govisual --transport http http://localhost:8080/mcp

Tools: get_last_error, list_requests, get_request, search_requests, get_stats, get_debug_context, replay_request, diff_replay, await_request, save_as_test, copy_as_curl, clear_requests.

Responses are token-aware: bounded lists and diagnostics, capped body excerpts, and sizes reported so the agent knows when to ask for more. The endpoint is loopback-only by default. Replay, diff, and generated curl commands require and are pinned to WithBaseURL(...).

Full guide: docs/claude-code.md.

Per-request capture

SQL queries

Wrap your database driver. Queries executed with the request's context land on that request's profile with duration, affected rows, and any error.

import (
	"database/sql"

	"github.com/doganarif/govisual/v2"
	_ "github.com/mattn/go-sqlite3" // your driver of choice
)

sql.Register("sqlite3+viz", govisual.WrapDriver(&sqlite3.SQLiteDriver{}))
db, err := sql.Open("sqlite3+viz", "app.db")
// then in a handler:
rows, err := db.QueryContext(r.Context(), "SELECT ...")

Enable govisual.WithProfiling(true) on the wrap call and the SQL panel shows up on the request drawer's Trace tab.

Outbound HTTP

Wrap your http.RoundTripper and outbound calls made with the inbound request's context are attached to it.

client := &http.Client{Transport: govisual.WrapTransport(nil)}
resp, err := client.Do(req.WithContext(r.Context()))

Application logs (slog)

Wrap your slog.Handler. Every record logged with the request's context also lands on that request's Logs tab, capped at 200 lines per request.

logger := slog.New(govisual.SlogHandler(slog.NewJSONHandler(os.Stdout, nil)))

// in a handler:
logger.InfoContext(r.Context(), "cache miss", "key", key)

Your base handler still receives every record unchanged.

Panics

Handler panics land on the captured request with the panic value and full stack, then the panic is re-raised so recovery middleware and net/http behave exactly as they would without govisual in the chain.

Custom events

govisual.Event(r.Context(), "cache miss", "key", key, "tier", "redis")

The event appears on the request's Logs tab and (with profiling on) inside the middleware trace.

Dashboard

http://localhost:8080/__viz by default. Customize with WithDashboardPath("/__debug").

• Inbox, Errors, Slow: filter captured requests by status and duration.
• Request drawer per request: Overview, Headers, Body, Trace (middleware, SQL, outbound HTTP), Logs, Performance (allocations, process heap, goroutines, GC, and a flame graph when profiling is on).
• Analytics: per-route request counts, p50/p95, error rates.
• Agents: recent MCP tool calls with their arguments, so you can watch a coding agent debug the app in real time.
• Environment: Go version, GOOS/GOARCH, memory stats, allowlisted env vars.
• Dark mode, toggled from the rail; choice persists via localStorage.
• Push updates (SSE): new requests appear on the dashboard as they land, not on a poll.

Storage backends

By default requests live in memory bounded by WithMaxRequests (100). For persistence, install one of the storage modules and pass its constructor to WithStore.

import (
	"github.com/doganarif/govisual/store/postgres"
	"github.com/doganarif/govisual/v2"
)

st, err := postgres.New(
	"postgres://user:pass@localhost:5432/dbname?sslmode=disable",
	"govisual_requests", // table, created and migrated automatically
	500,                 // capacity
)
if err != nil {
	log.Fatal(err)
}

handler := govisual.Wrap(mux, govisual.WithStore(st))

Same shape for the others:

redis.New("redis://localhost:6379/0", 500, 86400)         // capacity, TTL seconds
sqlite.New("requests.db", "govisual_requests", 500)
mongodb.New("mongodb://localhost:27017", "db", "coll", 500)

If your app already registers a SQLite driver, reuse your own *sql.DB:

db, _ := sql.Open("sqlite3", "app.db")
st, err := sqlite.NewWithDB(db, "govisual_requests", 500)

The v2 capture fields (logs, panic stacks, performance metrics) round-trip through every backend, including on tables that pre-date v2 (an ALTER TABLE runs on open).

OpenTelemetry

Tracing lives in its own module so the OTel SDK and gRPC stay out of builds that don't use them.

go get github.com/doganarif/govisual/telemetry

traced, shutdown, err := telemetry.Wrap(mux, telemetry.Config{
	ServiceName:    "my-service",
	ServiceVersion: "1.0.0",
	Endpoint:       "localhost:4317",
	Insecure:       true,
	Exporter:       "otlp", // or "stdout", "noop"
})
if err != nil {
	log.Fatal(err)
}
defer shutdown(context.Background())

// Wrap the mux (not the final handler) so govisual's dashboard stays untraced.
handler := govisual.Wrap(traced)

Configuration reference

…

Full option reference: docs/api-reference.md. Longer configuration guide: docs/configuration.md.

Dashboard security

The dashboard sees every captured request and response, so v2 defaults are conservative:

• Loopback-only unless WithAllowRemote(). Never expose a remote dashboard without WithBasicAuth, WithDashboardAuth, or equivalent authentication in an outer handler.
• Request replay is off unless WithReplayEnabled(true). Replays load a captured request by ID and are pinned to WithReplayBaseURL(...), or to a validated loopback dashboard origin when no base URL is configured. WithAllowRemote() requires an explicit replay base URL. Clients may edit the method, path, headers, and body, but cannot choose another destination host.
• System info is off unless WithSystemInfo(...). Environment variables are only exposed if you list them by name.
• Basic auth via WithBasicAuth(user, pass), or a custom check via WithDashboardAuth(func(*http.Request) bool).
• Sensitive headers (Authorization, Cookie, Set-Cookie, X-Api-Key, X-Auth-Token, X-Csrf-Token) are redacted at capture time. The header name is kept, the value is replaced.

Upgrading from v0.2.1

Three breaking changes:

1. Module path. github.com/doganarif/govisual becomes github.com/doganarif/govisual/v2. Update imports.
2. Storage options moved. WithMemoryStorage, WithPostgresStorage, WithSQLiteStorage, WithSQLiteStorageDB, WithRedisStorage, WithMongoDBStorage are gone. Install the backend module you want and pass the constructor to WithStore(...). See the Storage section above.
3. OpenTelemetry moved. WithOpenTelemetry, WithServiceName, WithServiceVersion, WithOTelEndpoint, WithOTelInsecure, WithOTelExporter are gone. Install github.com/doganarif/govisual/telemetry and wrap your mux with telemetry.Wrap(mux, telemetry.Config{...}) before handing it to govisual.Wrap.

Also worth noting:

• The dashboard is loopback-only now. Add WithAllowRemote() if you actually want it reachable over the network, and pair it with auth.
• Request replay is off unless you pass WithReplayEnabled(true).
• System info is off unless you pass WithSystemInfo(...).
• The captured RequestLog struct gained Logs, PanicStack, and Host. If you're reading raw JSON from the API, these are new optional fields; existing consumers keep working.

Examples

cd cmd/examples/basic         && go run main.go   # dashboard tour
cd cmd/examples/otel          && go run main.go   # OTel + telemetry module
cd cmd/examples/multistorage  && go run main.go   # postgres / redis / sqlite / mongo
cd cmd/examples/profiling     && go run main.go   # per-request profiling and flame graph
cd cmd/examples/tracing       && go run main.go   # middleware trace, Event, SQL capture

Documentation

• Quick start
• Configuration options
• Storage backends
• OpenTelemetry
• Claude Code / MCP
• API reference
• Troubleshooting
• FAQ

License

MIT.

Contributing

Pull requests welcome. See Contributing.

Built by Arif Dogan, production AI and backend engineer.

I help SaaS teams ship production AI features, fast backends, and reliable developer tools.

Work with me | Book a 30-min intro