Files
netbird/management/internals/modules/agentnetwork/handlers/policies_handler.go
T
Maycon Santos b416063bcc [management,proxy] Agent network: per-account LLM gateway (policy, metering, multi-provider) (#6555)
* [agent-network] Shared proto, OpenAPI schema, and generated types

* [agent-network] Management: store, manager, synthesizer, policy engine, provider catalog, HTTP/gRPC API

Adds the account-scoped agent-network module: provider/policy/budget CRUD and
store, the reverse-proxy service synthesizer, policy selection + limit
enforcement, the provider catalog (incl. Vertex AI and AWS Bedrock entries),
and the management HTTP + proxy gRPC surfaces.

* [management] Fix agent-network proxy-peer fan-out on affected-peer recompute

The affected-peers resolver loaded only persisted reverse-proxy services, but
agent-network services are synthesized on demand and never persisted. As a
result the embedded proxy peer was never folded into the affected set when a
client's group changed, so the proxy received no network-map update for a newly
authorised client and rejected its handshake until a full resync (restart).

loadProxyServices now merges the synthesized agent-network services (injected
via a registration hook to avoid an import cycle), so proxy peers learn newly
authorised clients immediately.

* [proxy] Reverse-proxy middleware framework, chain, and request plumbing

The per-target middleware chain (slots, dispatcher, mutation gate, metadata
merger), body capture, access-log terminal sink, and the proxy wiring that
builds + runs chains for synthesized agent-network services.

* [proxy] LLM parsers, pricing, and builtin middlewares (OpenAI, Anthropic, Vertex AI, AWS Bedrock)

Request/response parsers and SSE/event-stream metering, the embedded pricing
table, and the builtin middleware set: request parser, router, policy
limit-check/record, cost meter, guardrail, identity inject, response parser.
Includes the path-routed providers — Google Vertex AI (keyfile:: service-account
OAuth minting) and AWS Bedrock (bearer auth, invoke/converse/streaming, optional
/bedrock prefix) — plus the Models allowlist and unmeterable-publisher deny.

* [proxy] IPv6 in-place apply and TCP accept-loop hardening on netstack listeners

* [agent-network] End-to-end test suite, module docs, and deployment preset

* [agent-network] Fix codespell typos and exclude false positives

- labelgen word pool: vermillion -> vermilion, racoon -> raccoon.
- codespell ignore list: add flate (Go compress/flate package), recordin
  (a test-local identifier), and unparseable (a valid alternative spelling used
  consistently across identifiers + a metadata-value constant).

* [management] Set LastSeen on injected proxy peer in realstack test (MySQL strict-mode)

The injected embedded proxy peer had a PeerStatus with a zero LastSeen, which
serializes to '0000-00-00' and is rejected by MySQL in strict mode (SQLite
tolerates it). Set LastSeen to a valid time so SaveAccount succeeds on both
engines.

* [agent-network] Remove e2e shell-script suite from this branch

The end-to-end shell scripts under scripts/e2e/ are maintained in a separate
testing suite and are not part of this change set.

* [agent-network] Polish module docs: remove internal review scaffolding, fix links, verify diagrams

Strip PR-review framing, commit references, absolute paths, and stale internal
references from the agent-network module docs; fix broken relative links; verify
all diagrams against the current architecture. Remove the internal AI-reviewer
prompt file.

* [management] Refine session expiration handling to support 3-state encoding for SSO deadlines

* [agent-network] Relocate agentnetwork package to internals/modules

Move management/server/agentnetwork (and its catalog/, labelgen/, types/
subpackages) to management/internals/modules/agentnetwork, alongside the
reverse-proxy module, and rewrite all importers. Pure relocation: package names,
the synthesizer + affectedpeers registration hook, and store access (shared
store.Store) are unchanged, so no import cycle is introduced (affectedpeers
still depends only on the agentnetwork/types leaf).

* [agent-network] Co-locate HTTP handlers in the module (RegisterEndpoints)

Move the agent-network HTTP handlers from server/http/handlers/agentnetwork into
the module at internals/modules/agentnetwork/handlers (package handlers) and
rename the entrypoint AddEndpoints -> RegisterEndpoints, matching the
reverse-proxy module convention. Wiring in http/handler.go updated accordingly.
2026-06-27 13:41:00 +02:00

229 lines
7.4 KiB
Go

package handlers
import (
"encoding/json"
"net/http"
"strings"
"github.com/gorilla/mux"
"github.com/netbirdio/netbird/management/internals/modules/agentnetwork/types"
nbcontext "github.com/netbirdio/netbird/management/server/context"
"github.com/netbirdio/netbird/shared/management/http/api"
"github.com/netbirdio/netbird/shared/management/http/util"
"github.com/netbirdio/netbird/shared/management/status"
)
// minWindowSeconds is the floor enforced on enabled token / budget
// limit windows. One minute is short enough for fine-grained burst
// control without producing untenable consumption-row volume at scale.
const minWindowSeconds int64 = 60
// addPolicyEndpoints registers all Agent Network policy routes on the
// shared handler.
func (h *handler) addPolicyEndpoints(router *mux.Router) {
router.HandleFunc("/agent-network/policies", h.getAllPolicies).Methods("GET", "OPTIONS")
router.HandleFunc("/agent-network/policies", h.createPolicy).Methods("POST", "OPTIONS")
router.HandleFunc("/agent-network/policies/{policyId}", h.getPolicy).Methods("GET", "OPTIONS")
router.HandleFunc("/agent-network/policies/{policyId}", h.updatePolicy).Methods("PUT", "OPTIONS")
router.HandleFunc("/agent-network/policies/{policyId}", h.deletePolicy).Methods("DELETE", "OPTIONS")
}
func (h *handler) getAllPolicies(w http.ResponseWriter, r *http.Request) {
userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
policies, err := h.manager.GetAllPolicies(r.Context(), userAuth.AccountId, userAuth.UserId)
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
out := make([]*api.AgentNetworkPolicy, 0, len(policies))
for _, p := range policies {
out = append(out, p.ToAPIResponse())
}
util.WriteJSONObject(r.Context(), w, out)
}
func (h *handler) getPolicy(w http.ResponseWriter, r *http.Request) {
userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
policyID := mux.Vars(r)["policyId"]
if policyID == "" {
util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "policy ID is required"), w)
return
}
policy, err := h.manager.GetPolicy(r.Context(), userAuth.AccountId, userAuth.UserId, policyID)
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
util.WriteJSONObject(r.Context(), w, policy.ToAPIResponse())
}
func (h *handler) createPolicy(w http.ResponseWriter, r *http.Request) {
userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
var req api.AgentNetworkPolicyRequest
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
util.WriteErrorResponse("couldn't parse JSON request", http.StatusBadRequest, w)
return
}
if err := validatePolicy(&req); err != nil {
util.WriteError(r.Context(), err, w)
return
}
policy := types.NewPolicy(userAuth.AccountId)
policy.FromAPIRequest(&req)
created, err := h.manager.CreatePolicy(r.Context(), userAuth.UserId, policy)
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
util.WriteJSONObject(r.Context(), w, created.ToAPIResponse())
}
func (h *handler) updatePolicy(w http.ResponseWriter, r *http.Request) {
userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
policyID := mux.Vars(r)["policyId"]
if policyID == "" {
util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "policy ID is required"), w)
return
}
var req api.AgentNetworkPolicyRequest
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
util.WriteErrorResponse("couldn't parse JSON request", http.StatusBadRequest, w)
return
}
if err := validatePolicy(&req); err != nil {
util.WriteError(r.Context(), err, w)
return
}
policy := &types.Policy{
ID: policyID,
AccountID: userAuth.AccountId,
}
policy.FromAPIRequest(&req)
updated, err := h.manager.UpdatePolicy(r.Context(), userAuth.UserId, policy)
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
util.WriteJSONObject(r.Context(), w, updated.ToAPIResponse())
}
func (h *handler) deletePolicy(w http.ResponseWriter, r *http.Request) {
userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
if err != nil {
util.WriteError(r.Context(), err, w)
return
}
policyID := mux.Vars(r)["policyId"]
if policyID == "" {
util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "policy ID is required"), w)
return
}
if err := h.manager.DeletePolicy(r.Context(), userAuth.AccountId, userAuth.UserId, policyID); err != nil {
util.WriteError(r.Context(), err, w)
return
}
util.WriteJSONObject(r.Context(), w, util.EmptyObject{})
}
func validatePolicy(req *api.AgentNetworkPolicyRequest) error {
if strings.TrimSpace(req.Name) == "" {
return status.Errorf(status.InvalidArgument, "name is required")
}
if len(req.SourceGroups) == 0 {
return status.Errorf(status.InvalidArgument, "source_groups must contain at least one group id")
}
for _, id := range req.SourceGroups {
if strings.TrimSpace(id) == "" {
return status.Errorf(status.InvalidArgument, "source_groups must not contain empty entries")
}
}
if len(req.DestinationProviderIds) == 0 {
return status.Errorf(status.InvalidArgument, "destination_provider_ids must contain at least one provider id")
}
for _, id := range req.DestinationProviderIds {
if strings.TrimSpace(id) == "" {
return status.Errorf(status.InvalidArgument, "destination_provider_ids must not contain empty entries")
}
}
if req.GuardrailIds != nil {
for _, id := range *req.GuardrailIds {
if strings.TrimSpace(id) == "" {
return status.Errorf(status.InvalidArgument, "guardrail_ids must not contain empty entries")
}
}
}
if req.Limits != nil {
if err := validatePolicyLimits(*req.Limits); err != nil {
return err
}
}
return nil
}
func validatePolicyLimits(l api.AgentNetworkPolicyLimits) error {
if l.TokenLimit.Enabled {
if l.TokenLimit.WindowSeconds < minWindowSeconds {
return status.Errorf(status.InvalidArgument, "limits.token_limit.window_seconds must be at least %d (one minute) when enabled", minWindowSeconds)
}
if l.TokenLimit.GroupCap < 0 {
return status.Errorf(status.InvalidArgument, "limits.token_limit.group_cap must not be negative")
}
if l.TokenLimit.UserCap < 0 {
return status.Errorf(status.InvalidArgument, "limits.token_limit.user_cap must not be negative")
}
if l.TokenLimit.GroupCap == 0 && l.TokenLimit.UserCap == 0 {
return status.Errorf(status.InvalidArgument, "limits.token_limit requires group_cap or user_cap to be greater than zero when enabled")
}
}
if l.BudgetLimit.Enabled {
if l.BudgetLimit.WindowSeconds < minWindowSeconds {
return status.Errorf(status.InvalidArgument, "limits.budget_limit.window_seconds must be at least %d (one minute) when enabled", minWindowSeconds)
}
if l.BudgetLimit.GroupCapUsd < 0 {
return status.Errorf(status.InvalidArgument, "limits.budget_limit.group_cap_usd must not be negative")
}
if l.BudgetLimit.UserCapUsd < 0 {
return status.Errorf(status.InvalidArgument, "limits.budget_limit.user_cap_usd must not be negative")
}
if l.BudgetLimit.GroupCapUsd == 0 && l.BudgetLimit.UserCapUsd == 0 {
return status.Errorf(status.InvalidArgument, "limits.budget_limit requires group_cap_usd or user_cap_usd to be greater than zero when enabled")
}
}
return nil
}