checker: add honeypot-path collector and rules

Probes 20 known-bad paths (/.env, /.git/config, /actuator/env, etc.)
that CT-log scanners hit immediately after a new certificate is issued.
Critical credential/source-leak paths raise StatusCrit; other exposed
paths raise StatusWarn; 401/403 responses raise StatusInfo.

Fixes: #1

checker/collector.go

@@ -6,6 +6,11 @@ package checker
 
 import (
 	"context"
+	"crypto/tls"
+	"io"
+	"net"
+	"net/http"
+	"net/url"
 	"time"
 )
 
@@ -34,3 +39,54 @@ type Collector interface {
 	Key() string
 	Collect(ctx context.Context, t Target) (any, error)
 }
+
+// PathProbe is the common result of a single HTTPS path probe. It is
+// embedded by collector-specific probe types that may add extra fields
+// (e.g. HoneypotProbe adds Critical).
+type PathProbe struct {
+	URL        string `json:"url"`
+	StatusCode int    `json:"status_code,omitempty"`
+	Bytes      int    `json:"bytes,omitempty"`
+	Error      string `json:"error,omitempty"`
+}
+
+// fetchHTTPSPath issues a single GET against the given path using client,
+// reads up to limit bytes (just to measure size), and returns a PathProbe.
+func fetchHTTPSPath(ctx context.Context, client *http.Client, host, path, ua string, limit int64) PathProbe {
+	u := (&url.URL{Scheme: "https", Host: host, Path: path}).String()
+	probe := PathProbe{URL: u}
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, u, nil)
+	if err != nil {
+		probe.Error = err.Error()
+		return probe
+	}
+	req.Header.Set("User-Agent", ua)
+	resp, err := client.Do(req)
+	if err != nil {
+		probe.Error = err.Error()
+		return probe
+	}
+	defer resp.Body.Close()
+	probe.StatusCode = resp.StatusCode
+	n, _ := io.Copy(io.Discard, io.LimitReader(resp.Body, limit))
+	probe.Bytes = int(n)
+	return probe
+}
+
+// newPinnedHTTPSTransport returns an http.Transport that dials every request
+// to ip:443 and presents host as the TLS ServerName. The caller must defer
+// the returned cleanup func to drain idle connections.
+func newPinnedHTTPSTransport(ip, host string, timeout time.Duration) (*http.Transport, func()) {
+	addr := net.JoinHostPort(ip, "443")
+	dialer := &net.Dialer{Timeout: timeout}
+	t := &http.Transport{
+		DialContext: func(ctx context.Context, network, _ string) (net.Conn, error) {
+			return dialer.DialContext(ctx, network, addr)
+		},
+		TLSClientConfig:       &tls.Config{ServerName: host},
+		TLSHandshakeTimeout:   timeout,
+		ResponseHeaderTimeout: timeout,
+		DisableKeepAlives:     true,
+	}
+	return t, t.CloseIdleConnections
+}

checker/collector_honeypot.go

@@ -0,0 +1,109 @@
+// This file is part of the happyDomain (R) project.
+// Copyright (c) 2020-2026 happyDomain
+// Authors: Pierre-Olivier Mercier, et al.
+
+package checker
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"sync"
+)
+
+const ObservationKeyHoneypot = "honeypot"
+
+// honeypotPaths is the curated list of paths that internet scanners probe
+// immediately after a new TLS certificate appears in CT logs. Any path
+// returning a non-404/410 response is worth reporting to the site owner.
+//
+// /.well-known/security.txt is intentionally absent — it is already
+// evaluated by securityTxtRule via the wellknown collector.
+var honeypotPaths = []struct {
+	path     string
+	critical bool // true → leak of credentials or source code
+}{
+	// Credentials / source leaks — critical
+	{"/.env", true},
+	{"/.git/config", true},
+	{"/.vscode/sftp.json", true},
+	{"/.DS_Store", true},
+	// Debug / metrics endpoints
+	{"/actuator/env", true},
+	{"/server-status", false},
+	{"/debug/default/view", false},
+	{"/trace.axd", false},
+	{"/@vite/env", false},
+	{"/info.php", false},
+	// Admin panels
+	{"/console/", false},
+	{"/server", false},
+	{"/login.action", false},
+	{"/telescope/requests", false},
+	// APIs / registries
+	{"/v2/_catalog", false},
+	// CMS entry points
+	{"/xmlrpc.php", false},
+	{"/wp-json/", false},
+	{"/?rest_route=/wp/v2/users/", false},
+	// Config dumps
+	{"/config.json", false},
+	{"/version", false},
+}
+
+// HoneypotData is the observation written under ObservationKeyHoneypot.
+type HoneypotData struct {
+	Probes map[string]HoneypotProbe `json:"probes"`
+}
+
+// HoneypotProbe is the outcome of a single path probe.
+type HoneypotProbe struct {
+	PathProbe
+	Critical bool `json:"critical,omitempty"`
+}
+
+// honeypotCollector probes known-bad paths on the first HTTPS IP without
+// following redirects. A redirect is itself a finding.
+type honeypotCollector struct{}
+
+func (honeypotCollector) Key() string { return ObservationKeyHoneypot }
+
+func (honeypotCollector) Collect(ctx context.Context, t Target) (any, error) {
+	if len(t.IPs) == 0 {
+		return nil, fmt.Errorf("no IPs to probe")
+	}
+	transport, cleanup := newPinnedHTTPSTransport(t.IPs[0], t.Host, t.Timeout)
+	defer cleanup()
+	// No redirect following: a redirect to /login is itself a finding.
+	client := &http.Client{
+		Transport:     transport,
+		CheckRedirect: func(*http.Request, []*http.Request) error { return http.ErrUseLastResponse },
+	}
+
+	out := HoneypotData{Probes: make(map[string]HoneypotProbe, len(honeypotPaths))}
+	var mu sync.Mutex
+	var wg sync.WaitGroup
+	sem := make(chan struct{}, MaxConcurrentProbes)
+	for _, hp := range honeypotPaths {
+		wg.Add(1)
+		sem <- struct{}{}
+		go func(hp struct {
+			path     string
+			critical bool
+		}) {
+			defer wg.Done()
+			defer func() { <-sem }()
+			probe := HoneypotProbe{
+				PathProbe: fetchHTTPSPath(ctx, client, t.Host, hp.path, t.UserAgent, 4<<10),
+				Critical:  hp.critical,
+			}
+			mu.Lock()
+			out.Probes[hp.path] = probe
+			mu.Unlock()
+		}(hp)
+	}
+	wg.Wait()
+	return &out, nil
+}
+
+func init() { RegisterCollector(honeypotCollector{}) }

checker/collector_wellknown.go

@@ -6,12 +6,8 @@ package checker
 
 import (
 	"context"
-	"crypto/tls"
 	"fmt"
-	"io"
-	"net"
 	"net/http"
-	"net/url"
 )
 
 // ObservationKeyWellKnown is the Extensions[] key under which
@@ -28,12 +24,7 @@ type WellKnownData struct {
 }
 
 // WellKnownProbe is a single (URI → outcome) entry.
-type WellKnownProbe struct {
-	URL        string `json:"url"`
-	StatusCode int    `json:"status_code,omitempty"`
-	Bytes      int    `json:"bytes,omitempty"`
-	Error      string `json:"error,omitempty"`
-}
+type WellKnownProbe = PathProbe
 
 // wellknownCollector probes a small, fixed set of standardised URIs
 // served at the apex of the host. Today it covers:
@@ -51,49 +42,16 @@ func (wellknownCollector) Collect(ctx context.Context, t Target) (any, error) {
 	if len(t.IPs) == 0 {
 		return nil, fmt.Errorf("no IPs to probe")
 	}
-	addr := net.JoinHostPort(t.IPs[0], "443")
-	dialer := &net.Dialer{Timeout: t.Timeout}
-	transport := &http.Transport{
-		DialContext: func(ctx context.Context, network, _ string) (net.Conn, error) {
-			return dialer.DialContext(ctx, network, addr)
-		},
-		TLSClientConfig:       &tls.Config{ServerName: t.Host},
-		TLSHandshakeTimeout:   t.Timeout,
-		ResponseHeaderTimeout: t.Timeout,
-		DisableKeepAlives:     true,
-	}
-	defer transport.CloseIdleConnections()
+	transport, cleanup := newPinnedHTTPSTransport(t.IPs[0], t.Host, t.Timeout)
+	defer cleanup()
 	client := &http.Client{Transport: transport}
 
 	uris := []string{"/.well-known/security.txt", "/robots.txt"}
 	out := WellKnownData{URIs: make(map[string]WellKnownProbe, len(uris))}
 	for _, path := range uris {
-		out.URIs[path] = fetchOne(ctx, client, t.Host, path, t.UserAgent)
+		out.URIs[path] = fetchHTTPSPath(ctx, client, t.Host, path, t.UserAgent, 64<<10)
 	}
 	return &out, nil
 }
 
-func fetchOne(ctx context.Context, client *http.Client, host, path, ua string) WellKnownProbe {
-	u := (&url.URL{Scheme: "https", Host: host, Path: path}).String()
-	probe := WellKnownProbe{URL: u}
-	req, err := http.NewRequestWithContext(ctx, http.MethodGet, u, nil)
-	if err != nil {
-		probe.Error = err.Error()
-		return probe
-	}
-	req.Header.Set("User-Agent", ua)
-	resp, err := client.Do(req)
-	if err != nil {
-		probe.Error = err.Error()
-		return probe
-	}
-	defer resp.Body.Close()
-	probe.StatusCode = resp.StatusCode
-	// Cap the read so a misconfigured server can't pull megabytes for a
-	// "did this exist?" probe.
-	body, _ := io.ReadAll(io.LimitReader(resp.Body, 64<<10))
-	probe.Bytes = len(body)
-	return probe
-}
-
 func init() { RegisterCollector(wellknownCollector{}) }

checker/rules_honeypot.go

@@ -0,0 +1,81 @@
+// This file is part of the happyDomain (R) project.
+// Copyright (c) 2020-2026 happyDomain
+// Authors: Pierre-Olivier Mercier, et al.
+
+package checker
+
+import (
+	"context"
+	"fmt"
+
+	sdk "git.happydns.org/checker-sdk-go/checker"
+)
+
+func init() { RegisterRule(&honeypotRule{}) }
+
+// honeypotRule reports any honeypot path that returned a non-404/410
+// response, signalling a potentially exposed sensitive endpoint.
+type honeypotRule struct{}
+
+func (r *honeypotRule) Name() string { return "http.honeypot" }
+func (r *honeypotRule) Description() string {
+	return "Reports sensitive paths (/.env, /.git/config, /actuator/env, …) that are reachable from the internet."
+}
+
+func (r *honeypotRule) Evaluate(ctx context.Context, obs sdk.ObservationGetter, _ sdk.CheckerOptions) []sdk.CheckState {
+	data, errSt := loadHTTPData(ctx, obs)
+	if errSt != nil {
+		return []sdk.CheckState{*errSt}
+	}
+	hp, ok, err := LoadExtension[HoneypotData](data, ObservationKeyHoneypot)
+	if err != nil {
+		return []sdk.CheckState{{Status: sdk.StatusError, Code: "http.honeypot.decode_error", Message: err.Error()}}
+	}
+	if !ok {
+		return []sdk.CheckState{unknownState("http.honeypot.no_data", "Honeypot collector did not run.")}
+	}
+
+	var states []sdk.CheckState
+	successfulProbes := 0
+	for path, probe := range hp.Probes {
+		if probe.Error != "" || probe.StatusCode == 0 {
+			continue
+		}
+		successfulProbes++
+
+		st := sdk.CheckState{
+			Subject: path,
+			Meta:    map[string]any{"url": probe.URL, "status_code": probe.StatusCode},
+		}
+		switch {
+		case probe.StatusCode == 200 || (probe.StatusCode >= 301 && probe.StatusCode <= 308):
+			st.Status = sdk.StatusWarn
+			st.Code = "http.honeypot.exposed"
+			st.Message = fmt.Sprintf("%s is accessible (HTTP %d, %d bytes).", path, probe.StatusCode, probe.Bytes)
+			if probe.Critical {
+				st.Status = sdk.StatusCrit
+				st.Code = "http.honeypot.critical_exposed"
+			}
+		case probe.StatusCode == 401 || probe.StatusCode == 403:
+			st.Status = sdk.StatusInfo
+			st.Code = "http.honeypot.protected"
+			st.Message = fmt.Sprintf("%s exists but is access-controlled (HTTP %d).", path, probe.StatusCode)
+		default:
+			continue
+		}
+		states = append(states, st)
+	}
+
+	if successfulProbes == 0 {
+		return []sdk.CheckState{unknownState("http.honeypot.no_response", "All honeypot probes failed or timed out.")}
+	}
+	if len(states) == 0 {
+		return []sdk.CheckState{{
+			Status:  sdk.StatusOK,
+			Code:    "http.honeypot.clean",
+			Subject: data.Domain,
+			Message: "No sensitive honeypot paths are reachable.",
+		}}
+	}
+	return states
+}

checker/rules_honeypot_test.go

@@ -0,0 +1,147 @@
+// This file is part of the happyDomain (R) project.
+// Copyright (c) 2020-2026 happyDomain
+// Authors: Pierre-Olivier Mercier, et al.
+
+package checker
+
+import (
+	"encoding/json"
+	"testing"
+
+	sdk "git.happydns.org/checker-sdk-go/checker"
+)
+
+func honeypotData(t *testing.T, probes map[string]HoneypotProbe) map[string]json.RawMessage {
+	t.Helper()
+	raw, err := json.Marshal(HoneypotData{Probes: probes})
+	if err != nil {
+		t.Fatalf("marshal: %v", err)
+	}
+	return map[string]json.RawMessage{ObservationKeyHoneypot: raw}
+}
+
+func TestHoneypotRule_Clean(t *testing.T) {
+	probes := map[string]HoneypotProbe{
+		"/.env":         {PathProbe: PathProbe{StatusCode: 404}},
+		"/.git/config":  {PathProbe: PathProbe{StatusCode: 404}},
+		"/actuator/env": {PathProbe: PathProbe{StatusCode: 404}},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusOK)
+	if !hasCode(states, "http.honeypot.clean") {
+		t.Errorf("expected clean, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_CriticalExposed(t *testing.T) {
+	probes := map[string]HoneypotProbe{
+		"/.env": {PathProbe: PathProbe{StatusCode: 200, Bytes: 512}, Critical: true},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusCrit)
+	if !hasCode(states, "http.honeypot.critical_exposed") {
+		t.Errorf("expected critical_exposed, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_WarnExposed(t *testing.T) {
+	probes := map[string]HoneypotProbe{
+		"/server-status": {PathProbe: PathProbe{StatusCode: 200, Bytes: 1024}},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusWarn)
+	if !hasCode(states, "http.honeypot.exposed") {
+		t.Errorf("expected exposed, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_Redirect(t *testing.T) {
+	probes := map[string]HoneypotProbe{
+		"/console/": {PathProbe: PathProbe{StatusCode: 301}},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusWarn)
+	if !hasCode(states, "http.honeypot.exposed") {
+		t.Errorf("expected exposed for redirect, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_Protected(t *testing.T) {
+	probes := map[string]HoneypotProbe{
+		"/.git/config": {PathProbe: PathProbe{StatusCode: 403}, Critical: true},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusInfo)
+	if !hasCode(states, "http.honeypot.protected") {
+		t.Errorf("expected protected, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_NoCollectorData(t *testing.T) {
+	data := &HTTPData{
+		Domain: "example.test",
+		Probes: []HTTPProbe{httpsProbe("a:443")},
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusUnknown)
+	if !hasCode(states, "http.honeypot.no_data") {
+		t.Errorf("expected no_data, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_DecodeError(t *testing.T) {
+	data := &HTTPData{
+		Domain: "example.test",
+		Probes: []HTTPProbe{httpsProbe("a:443")},
+		Extensions: map[string]json.RawMessage{
+			ObservationKeyHoneypot: json.RawMessage(`"not an object"`),
+		},
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	if states[0].Status != sdk.StatusError || states[0].Code != "http.honeypot.decode_error" {
+		t.Errorf("expected decode_error, got %+v", states)
+	}
+}
+
+func TestHoneypotRule_IgnoresErrors(t *testing.T) {
+	// All-error probes must return Unknown, not OK, to avoid false-green.
+	probes := map[string]HoneypotProbe{
+		"/.env":     {PathProbe: PathProbe{Error: "connection refused"}},
+		"/wp-json/": {},
+	}
+	data := &HTTPData{
+		Domain:     "example.test",
+		Probes:     []HTTPProbe{httpsProbe("a:443")},
+		Extensions: honeypotData(t, probes),
+	}
+	states := runRule(t, &honeypotRule{}, data, nil)
+	mustStatus(t, states, sdk.StatusUnknown)
+	if !hasCode(states, "http.honeypot.no_response") {
+		t.Errorf("expected no_response when all probes errored, got %+v", states)
+	}
+}