Files
netbird/proxy/internal/middleware/builtin/llm_router/factory.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

107 lines
4.5 KiB
Go

package llm_router
import (
"bytes"
"encoding/json"
"fmt"
"strings"
"github.com/netbirdio/netbird/proxy/internal/middleware"
"github.com/netbirdio/netbird/proxy/internal/middleware/builtin"
)
// ProviderRoute describes one upstream LLM provider the router can
// hand a request to. Models lists the model identifiers the provider
// claims; UpstreamScheme + UpstreamHost replace the synth target's
// placeholder URL on a match. UpstreamPath is the path component of
// the configured upstream URL — the router uses it to disambiguate
// providers that claim the same model: when more than one provider
// matches the model, the route whose UpstreamPath is a prefix of the
// incoming request path is preferred (longest match wins, empty path
// is the catchall). AuthHeaderName + AuthHeaderValue are the
// per-provider credential the router injects after stripping the
// vendor auth headers from the inbound request.
//
// AllowedGroupIDs is the union of source-group IDs across every
// enabled policy that authorises this provider. The router treats it
// as a hard filter: a route whose AllowedGroupIDs has no intersection
// with the caller's UserGroups is removed from the candidate list
// before the path-prefix tiebreak. A route with empty AllowedGroupIDs
// is unreachable; the synthesiser only emits policy-bound routes.
type ProviderRoute struct {
ID string `json:"id"`
// Vendor is the parser surface this provider speaks ("openai",
// "anthropic", …), matching the llm.provider value llm_request_parser
// emits from the request. When set, the router keeps a vendor-tagged
// request on a same-vendor route so catch-all gateways of a different
// vendor can't swallow it. Empty disables vendor filtering for this
// route.
Vendor string `json:"vendor,omitempty"`
Models []string `json:"models"`
UpstreamScheme string `json:"upstream_scheme"`
UpstreamHost string `json:"upstream_host"`
UpstreamPath string `json:"upstream_path,omitempty"`
AuthHeaderName string `json:"auth_header_name"`
AuthHeaderValue string `json:"auth_header_value"`
AllowedGroupIDs []string `json:"allowed_group_ids"`
// Vertex marks a Google Vertex AI provider. Vertex requests carry the
// model in the URL path, so the router selects this route by path
// (isVertexPath) and bypasses the model/vendor table entirely.
Vertex bool `json:"vertex,omitempty"`
// Bedrock marks an AWS Bedrock provider. Bedrock requests carry the model
// in the URL path (/model/{id}/{action}), so the router selects this route
// by path (isBedrockPath) and bypasses the model/vendor table; auth is the
// static AuthHeaderValue bearer token (no token minting).
Bedrock bool `json:"bedrock,omitempty"`
// GCPServiceAccountKeyB64 is a base64-encoded GCP service-account JSON
// key. When set, the router mints + refreshes a short-lived OAuth2 access
// token from it at request time and injects it as the auth header value
// (instead of the static AuthHeaderValue) — so the gateway holds a durable
// Vertex credential rather than a 1-hour token.
GCPServiceAccountKeyB64 string `json:"gcp_sa_key_b64,omitempty"`
}
// Config is the on-wire configuration accepted by the factory. An
// empty Providers slice yields a router that denies every request as
// not-routable; the synthesiser is responsible for stamping the
// account's enabled providers into this slice.
type Config struct {
Providers []ProviderRoute `json:"providers"`
}
// Factory builds llm_router instances from raw config bytes.
type Factory struct{}
// ID returns the registry identifier.
func (Factory) ID() string { return ID }
// New constructs a middleware instance. Empty, null, and {} configs
// yield a router with an empty Providers slice — every request denies
// with model_not_routable. Non-empty payloads must parse cleanly so
// misconfigurations surface at chain build time.
func (Factory) New(rawConfig []byte) (middleware.Middleware, error) {
cfg := Config{}
if !isEmptyJSON(rawConfig) {
if err := json.Unmarshal(rawConfig, &cfg); err != nil {
return nil, fmt.Errorf("decode config: %w", err)
}
}
return New(cfg), nil
}
// isEmptyJSON reports whether the payload is whitespace, null, or an
// empty object/array. The caller skips Unmarshal in that case so the
// zero-value Config flows through unchanged.
func isEmptyJSON(raw []byte) bool {
trimmed := strings.TrimSpace(string(bytes.TrimSpace(raw)))
switch trimmed {
case "", "null", "{}", "[]":
return true
}
return false
}
func init() {
builtin.Register(Factory{})
}