// Package executor provides runtime execution capabilities for various AI service providers. // This file implements the Antigravity executor that proxies requests to the antigravity // upstream using OAuth credentials. package executor import ( "bytes" "context" "crypto/sha256" "crypto/tls" "encoding/hex" "encoding/json" "fmt" "net/http" "strings" "time" "github.com/router-for-me/CLIProxyAPI/v7/internal/cache" "github.com/router-for-me/CLIProxyAPI/v7/internal/config" "github.com/router-for-me/CLIProxyAPI/v7/internal/runtime/executor/helps" internalsignature "github.com/router-for-me/CLIProxyAPI/v7/internal/signature" antigravityclaude "github.com/router-for-me/CLIProxyAPI/v7/internal/translator/antigravity/claude" "github.com/router-for-me/CLIProxyAPI/v7/internal/util" cliproxyauth "github.com/router-for-me/CLIProxyAPI/v7/sdk/cliproxy/auth" "github.com/router-for-me/CLIProxyAPI/v7/sdk/proxyutil" sdktranslator "github.com/router-for-me/CLIProxyAPI/v7/sdk/translator" log "github.com/sirupsen/logrus" "github.com/tidwall/gjson" "github.com/tidwall/sjson" ) const ( antigravityBaseURLDaily = "https://daily-cloudcode-pa.googleapis.com" antigravitySandboxBaseURLDaily = "https://daily-cloudcode-pa.sandbox.googleapis.com" antigravityBaseURLProd = "https://cloudcode-pa.googleapis.com" antigravityCountTokensPath = "/v1internal:countTokens" antigravityStreamPath = "/v1internal:streamGenerateContent" antigravityGeneratePath = "/v1internal:generateContent" antigravityClientID = "1071006060591-tmhssin2h21lcre235vtolojh4g403ep.apps.googleusercontent.com" antigravityClientSecret = "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" antigravityAuthType = "antigravity" antigravityRequestTokenSafetyWindow = 5 * time.Minute antigravityCreditsHintRefreshInterval = 10 * time.Minute antigravityCreditsHintRefreshTimeout = 5 * time.Second antigravityShortQuotaCooldownThreshold = 5 * time.Minute antigravityInstantRetryThreshold = 3 * time.Second // systemInstruction = "You are Antigravity, a powerful agentic AI coding assistant designed by the Google Deepmind team working on Advanced Agentic Coding.You are pair programming with a USER to solve their coding task. The task may require creating a new codebase, modifying or debugging an existing codebase, or simply answering a question.**Absolute paths only****Proactiveness**" ) // AntigravityExecutor proxies requests to the antigravity upstream. type AntigravityExecutor struct { cfg *config.Config } // NewAntigravityExecutor creates a new Antigravity executor instance. // // Parameters: // - cfg: The application configuration // // Returns: // - *AntigravityExecutor: A new Antigravity executor instance func NewAntigravityExecutor(cfg *config.Config) *AntigravityExecutor { return &AntigravityExecutor{cfg: cfg} } func (e *AntigravityExecutor) obfuscateSensitiveWords(payload []byte) []byte { if e == nil || e.cfg == nil || len(e.cfg.Antigravity.SensitiveWords) == 0 { return payload } matcher := helps.BuildSensitiveWordMatcher(e.cfg.Antigravity.SensitiveWords) return helps.ObfuscateSensitiveWordsInSystemInstruction(payload, matcher) } // Each Antigravity credential gets its own HTTP/1.1 connection pool. Sessions routed // to the same auth reuse that pool, while different OAuth identities never share a // TCP/TLS connection, matching the native client's one-credential process model. // The cache is bounded so pools cannot accumulate when keys churn. var ( antigravityBaseTransport = defaultAntigravityBaseTransport() antigravityTransports = helps.NewTransportCache[antigravityTransportKey](antigravityTransportCacheCapacity) ) const ( // antigravityTransportCacheCapacity caps how many Antigravity connection pools stay // alive. The bound exists only to stop entries from accumulating when keys churn, for // example when a credential's proxy is rotated through the management API or when an // SDK embedder supplies a freshly built base transport per request. // // It is sized for large deployments on purpose. An unused cache entry costs under 1 KB // and no goroutines, so capacity is close to free, whereas evicting a pool that is // still in active use forces the next request on that credential to redo the TCP + TLS // handshake and defeats the point of caching. Credential counts in the low thousands // are expected once Home-managed pools are included. // // Capacity is therefore NOT the lever for bounding memory: an idle pooled connection // costs roughly 38 KB plus three goroutines, and that total is driven by live traffic // and reclaimed by IdleConnTimeout. Shrinking this number does not save that memory, // it only causes pool thrashing. antigravityTransportCacheCapacity = 8192 // antigravityDefaultMaxIdleConnsPerHost sets the default number of idle connections // to retain per host per credential when connection pooling is enabled. // Matches Go's DefaultMaxIdleConnsPerHost (2) and the native Antigravity binary. antigravityDefaultMaxIdleConnsPerHost = 2 // antigravityMaxAllowedMaxIdleConnsPerHost is the hard upper bound on MaxIdleConnsPerHost (100). // Prevents unbounded connection pool expansion in multi-credential environments. antigravityMaxAllowedMaxIdleConnsPerHost = 100 // antigravityDefaultIdleConnTimeout is the default idle connection timeout (30 seconds). // Kept strictly far below Google Frontend (GFE / ESF) 240-second cutoff to prevent // client-side reuse of half-closed connections that cause connection resets. antigravityDefaultIdleConnTimeout = 30 * time.Second // antigravityMaxAllowedIdleConnTimeout is the hard upper bound on IdleConnTimeout (210 seconds). // Kept strictly below Google Frontend (GFE / ESF) 240.0-second HTTP/1.1 idle keep-alive cutoff // with a 30-second safety margin to eliminate timer race conditions. antigravityMaxAllowedIdleConnTimeout = 210 * time.Second // antigravityAnonymousTransportScope is the pool scope for auth objects that carry // no identity at all. Reaching it means the auth has no ID, no source path and no // token of any kind, so there is no credential to keep isolated and a single shared // pool is safe. Allocating a private pool per request instead would leak a // connection pool, and the goroutines managing it, on every call. antigravityAnonymousTransportScope = "anonymous" ) type antigravityPoolSettings struct { shortMode bool idleConnTimeout time.Duration maxIdleConnsPerHost int } func resolveAntigravityPoolSettings(cfg *config.Config) antigravityPoolSettings { // By default, upstream connection pooling is disabled (shortMode = true, maxIdleConnsPerHost = -1) // to prevent socket buildup and stale connection errors across rotating credentials. settings := antigravityPoolSettings{ shortMode: true, maxIdleConnsPerHost: -1, idleConnTimeout: 0, } if cfg == nil { return settings } // Pooling is active ONLY when explicitly enabled: true if cfg.Antigravity.ConnectionPool.Enabled == nil || !*cfg.Antigravity.ConnectionPool.Enabled { return settings } // Enabled is true: initialize default pool settings settings.shortMode = false settings.idleConnTimeout = antigravityDefaultIdleConnTimeout settings.maxIdleConnsPerHost = antigravityDefaultMaxIdleConnsPerHost rawTimeout := strings.TrimSpace(cfg.Antigravity.ConnectionPool.IdleConnTimeout) if rawTimeout != "" { d, err := time.ParseDuration(rawTimeout) if err != nil { log.Warnf("antigravity executor: invalid idle-conn-timeout %q: %v, using default %v", rawTimeout, err, antigravityDefaultIdleConnTimeout) } else { if d <= 0 { settings.shortMode = true settings.maxIdleConnsPerHost = -1 return settings } if d > antigravityMaxAllowedIdleConnTimeout { d = antigravityMaxAllowedIdleConnTimeout } settings.idleConnTimeout = d } } if cfg.Antigravity.ConnectionPool.MaxIdleConnsPerHost != nil { val := *cfg.Antigravity.ConnectionPool.MaxIdleConnsPerHost if val < 0 { settings.shortMode = true settings.maxIdleConnsPerHost = -1 return settings } if val > antigravityMaxAllowedMaxIdleConnsPerHost { val = antigravityMaxAllowedMaxIdleConnsPerHost } settings.maxIdleConnsPerHost = val } if settings.maxIdleConnsPerHost < 0 { settings.shortMode = true } return settings } // ResetAntigravityTransports purges all cached Antigravity connection pools and closes their idle connections. // Used during configuration hot-reloads to ensure updated pool parameters apply immediately. func ResetAntigravityTransports() { antigravityTransports.Purge() } // AntigravityTransportsLen reports the current number of cached Antigravity transports. func AntigravityTransportsLen() int { return antigravityTransports.Len() } // closeAntigravityAuthIdleTransports closes and removes all idle connections for the given auth. func closeAntigravityAuthIdleTransports(auth *cliproxyauth.Auth) { if auth == nil { return } scope := antigravityTransportScope(auth) if scope == "" || scope == antigravityAnonymousTransportScope { return } antigravityTransports.CloseMatching(func(key antigravityTransportKey) bool { return key.credential == scope }) } // antigravityTransportKey identifies one connection pool. At most one of proxy and // base is set: proxy for a credential-scoped proxy pool, base for a transport handed // in through the request context, and neither for a direct pool. // Resolved pool settings (shortMode, idleConnTimeout, maxIdleConnsPerHost) are included // in the key to prevent stale transport reuse across configuration hot-reloads under load. type antigravityTransportKey struct { credential string proxy string base *http.Transport shortMode bool idleConnTimeout time.Duration maxIdleConnsPerHost int } func defaultAntigravityBaseTransport() *http.Transport { if transport, ok := http.DefaultTransport.(*http.Transport); ok && transport != nil { return transport } return &http.Transport{} } func cloneTransportWithHTTP11(base *http.Transport, cfgs ...*config.Config) *http.Transport { if base == nil { return nil } clone := base.Clone() clone.ForceAttemptHTTP2 = false // Wipe TLSNextProto to prevent implicit HTTP/2 upgrade. clone.TLSNextProto = make(map[string]func(authority string, c *tls.Conn) http.RoundTripper) if clone.TLSClientConfig == nil { clone.TLSClientConfig = &tls.Config{} } else { clone.TLSClientConfig = clone.TLSClientConfig.Clone() } // Native Antigravity sends no ALPN extension. With HTTP/2 disabled above, // an empty NextProtos keeps the wire shape aligned while using HTTP/1.1. clone.TLSClientConfig.NextProtos = nil applyAntigravityPoolLimits(clone, cfgs...) return clone } // applyAntigravityPoolLimits configures connection pool parameters for Antigravity. // Default: IdleConnTimeout=30s, MaxIdleConnsPerHost=2. // IdleConnTimeout is strictly capped at 240s (antigravityMaxAllowedIdleConnTimeout) // matching the Google Frontend (GFE / ESF) HTTP/1.1 idle cutoff. // If short-lived connection mode is configured, MaxIdleConnsPerHost is set to -1 // so idle connections are closed immediately upon request completion. func applyAntigravityPoolLimits(transport *http.Transport, cfgs ...*config.Config) { if transport == nil { return } var cfg *config.Config if len(cfgs) > 0 { cfg = cfgs[0] } settings := resolveAntigravityPoolSettings(cfg) if settings.shortMode { transport.MaxIdleConnsPerHost = -1 transport.DisableKeepAlives = false transport.IdleConnTimeout = 0 return } // If the operator base transport already disabled pooling (< 0), honor it. if transport.MaxIdleConnsPerHost < 0 { return } // Go treats 0 as DefaultMaxIdleConnsPerHost (2). Ensure at least configured/default limit. if transport.MaxIdleConnsPerHost < settings.maxIdleConnsPerHost { transport.MaxIdleConnsPerHost = settings.maxIdleConnsPerHost } // MaxIdleConns caps the pool across all hosts. if transport.MaxIdleConns > 0 && transport.MaxIdleConns < transport.MaxIdleConnsPerHost { transport.MaxIdleConns = transport.MaxIdleConnsPerHost } // Apply IdleConnTimeout: // If the config explicitly specified a timeout, apply settings.idleConnTimeout. // If config did not specify a timeout: // - if base transport already has a timeout > 0, preserve it (capped at 240s). // - otherwise, apply default 30s. rawTimeout := "" if cfg != nil { rawTimeout = strings.TrimSpace(cfg.Antigravity.ConnectionPool.IdleConnTimeout) } if rawTimeout != "" { transport.IdleConnTimeout = settings.idleConnTimeout } else if transport.IdleConnTimeout == 0 || transport.IdleConnTimeout == 90*time.Second { transport.IdleConnTimeout = settings.idleConnTimeout } else if transport.IdleConnTimeout > antigravityMaxAllowedIdleConnTimeout { transport.IdleConnTimeout = antigravityMaxAllowedIdleConnTimeout } } // antigravityHTTP11Transport returns the HTTP/1.1 pool shared by every request that // uses the same credential and the same base transport. The base is either the // process default or a transport provided through the request context. func antigravityHTTP11Transport(auth *cliproxyauth.Auth, base *http.Transport, cfgs ...*config.Config) *http.Transport { if base == nil { return nil } var cfg *config.Config if len(cfgs) > 0 { cfg = cfgs[0] } settings := resolveAntigravityPoolSettings(cfg) key := antigravityTransportKey{ credential: antigravityTransportScope(auth), base: base, shortMode: settings.shortMode, idleConnTimeout: settings.idleConnTimeout, maxIdleConnsPerHost: settings.maxIdleConnsPerHost, } transport, errGet := antigravityTransports.Get(key, func() (*http.Transport, error) { return cloneTransportWithHTTP11(base, cfgs...), nil }) if errGet != nil { // Defensive only: the builder above cannot fail. Never return nil here, because a // nil Transport makes http.Client fall back to http.DefaultTransport, which // advertises h2 over ALPN and would break the Antigravity wire fingerprint. log.Debugf("antigravity executor: cache HTTP/1.1 transport failed: %v", errGet) return cloneTransportWithHTTP11(base, cfgs...) } return transport } // antigravityProxiedHTTP11Transport returns the credential-scoped HTTP/1.1 pool for // one proxy setting, or nil when the proxy setting cannot be turned into a // transport. Keying on the normalized proxy string rather than on a prebuilt // transport keeps one pool per credential and proxy instead of one per request. func antigravityProxiedHTTP11Transport(auth *cliproxyauth.Auth, proxyURL string, cfgs ...*config.Config) *http.Transport { proxyURL = strings.TrimSpace(proxyURL) if proxyURL == "" { return nil } var cfg *config.Config if len(cfgs) > 0 { cfg = cfgs[0] } settings := resolveAntigravityPoolSettings(cfg) key := antigravityTransportKey{ credential: antigravityTransportScope(auth), proxy: proxyURL, shortMode: settings.shortMode, idleConnTimeout: settings.idleConnTimeout, maxIdleConnsPerHost: settings.maxIdleConnsPerHost, } transport, errGet := antigravityTransports.Get(key, func() (*http.Transport, error) { base, _, errBuild := proxyutil.BuildHTTPTransport(proxyURL) if errBuild != nil { return nil, errBuild } if base == nil { return nil, fmt.Errorf("antigravity executor: proxy setting produced no transport") } return cloneTransportWithHTTP11(base, cfgs...), nil }) if errGet != nil { // The caller falls back to NewProxyAwareHTTPClient, which reports the failure // and applies the context transport fallback. return nil } return transport } // antigravityTransportScope returns the connection-pool scope for one credential. // Runtime auths always carry an ID. Incomplete auth objects, such as those built by // tests, plugins or SDK embedders, fall back to another stable credential marker so // they neither share a pool with an unrelated OAuth identity nor allocate a fresh // pool, and with it a fresh set of pool goroutines, on every single request. func antigravityTransportScope(auth *cliproxyauth.Auth) string { if auth == nil { return antigravityAnonymousTransportScope } if id := strings.TrimSpace(auth.ID); id != "" { return "id:" + id } if auth.Attributes != nil { if path := strings.TrimSpace(auth.Attributes[cliproxyauth.AttributePath]); path != "" { return "path:" + path } if source := strings.TrimSpace(auth.Attributes[cliproxyauth.AttributeSource]); source != "" { return "source:" + source } } // Fall back to the credential material itself. Auth.Label is deliberately not used: // it is documented as an optional human readable label for logging and carries no // uniqueness guarantee, so two different OAuth identities sharing one label would // wrongly share a TCP/TLS pool. // // The refresh token is preferred over the access token because it stays stable // across token rotation. Keying on the access token would move a credential to a new // pool on every refresh, and would also strand refresh requests themselves, which // run before any access token exists. if refresh := strings.TrimSpace(metaStringValue(auth.Metadata, "refresh_token")); refresh != "" { return antigravityCredentialScope("refresh:", refresh) } if access := strings.TrimSpace(metaStringValue(auth.Metadata, "access_token")); access != "" { return antigravityCredentialScope("token:", access) } return antigravityAnonymousTransportScope } // antigravityCredentialScope derives a pool scope from secret credential material. // Only a short digest is retained, and it is never logged, so a pool key cannot be // used to recover the credential it came from. func antigravityCredentialScope(prefix, secret string) string { digest := sha256.Sum256([]byte(secret)) return prefix + hex.EncodeToString(digest[:8]) } // newAntigravityHTTPClient creates an HTTP client specifically for Antigravity, // enforcing HTTP/1.1 by disabling HTTP/2 to match the native Antigravity client, which // negotiates TLS 1.3 without advertising an ALPN protocol and therefore never uses h2. // The underlying Transport is always shared so keep-alive connections survive across // requests instead of forcing a fresh TCP + TLS handshake every time. func newAntigravityHTTPClient(ctx context.Context, cfg *config.Config, auth *cliproxyauth.Auth, timeout time.Duration) *http.Client { // Native Antigravity reuses one transport across requests. Opt into a // credential-scoped proxy transport only here so other providers keep their // existing lifecycle and different OAuth identities remain isolated. if proxyURL := antigravityProxyURL(cfg, auth); proxyURL != "" { if transport := antigravityProxiedHTTP11Transport(auth, proxyURL, cfg); transport != nil { return &http.Client{Transport: transport, Timeout: timeout} } // Fall through so NewProxyAwareHTTPClient reports the failure and applies the // context transport fallback, preserving the previous behavior. } client := helps.NewProxyAwareHTTPClient(ctx, cfg, auth, timeout) // Direct requests share an HTTP/1.1 pool only within the selected credential. if client.Transport == nil { client.Transport = antigravityHTTP11Transport(auth, antigravityBaseTransport, cfg) return client } // Preserve a context-provided transport while forcing HTTP/1.1. The cache key // includes credential identity, so sharing the base does not share TLS pools. transport, ok := client.Transport.(*http.Transport) if !ok { // A RoundTripper that is not an *http.Transport owns its own protocol behavior. return client } if transport == nil { // A typed-nil *http.Transport still satisfies the interface nil check in // NewProxyAwareHTTPClient. Leaving it in place would make http.Client fall back // to http.DefaultTransport, which advertises h2 over ALPN and breaks the // Antigravity fingerprint, so substitute the process base transport. transport = antigravityBaseTransport } client.Transport = antigravityHTTP11Transport(auth, transport, cfg) return client } func antigravityProxyURL(cfg *config.Config, auth *cliproxyauth.Auth) string { if auth != nil { if proxyURL := strings.TrimSpace(auth.ProxyURL); proxyURL != "" { return proxyURL } } if cfg != nil { return strings.TrimSpace(cfg.ProxyURL) } return "" } func sanitizeAntigravityGeminiRequestSignatures(modelName string, rawJSON []byte) []byte { if !antigravityUsesReasoningReplayCache(modelName) { return rawJSON } rawJSON = internalsignature.SanitizeGeminiRequestThoughtSignatures(rawJSON, "request.contents") return normalizeAntigravityGeminiFunctionResponseRoles(rawJSON) } // ensureAntigravityGeminiLeadingUserContent prepends a synthetic empty user turn // after every contents rewrite, including reasoning replay. Claude targets are // left unchanged because the adapter rejects empty text parts. func ensureAntigravityGeminiLeadingUserContent(modelName string, payload []byte) []byte { if strings.Contains(strings.ToLower(modelName), "claude") { return payload } return helps.EnsureGeminiLeadingUserContent(payload, "request.contents") } // ensureAntigravityGeminiTrailingUserContent appends a synthetic empty user turn // if the final turn is a model turn. Claude targets are left unchanged because // the adapter rejects empty text parts. func ensureAntigravityGeminiTrailingUserContent(modelName string, payload []byte) []byte { if strings.Contains(strings.ToLower(modelName), "claude") { return payload } return helps.EnsureGeminiTrailingUserContent(payload, "request.contents") } // ensureAntigravityGeminiBoundaryUserContent normalizes both leading and trailing // turns for Gemini targets. Claude targets are left unchanged. func ensureAntigravityGeminiBoundaryUserContent(modelName string, payload []byte) []byte { if strings.Contains(strings.ToLower(modelName), "claude") { return payload } return helps.EnsureGeminiBoundaryUserContent(payload, "request.contents") } type antigravityContentEdit struct { index int64 path string start int end int replacement []byte } // normalizeAntigravityGeminiFunctionResponseRoles edits each response turn in // isolation, then splices all changed turns into the request with one body copy. // Applying SJSON once per field made large histories scale with history size // multiplied by the number of tool turns. func normalizeAntigravityGeminiFunctionResponseRoles(rawJSON []byte) []byte { rawJSON = repairAntigravityGeminiFunctionResponseNames(rawJSON) contents := util.GetGJSONBytesNoCopy(rawJSON, "request.contents") if !contents.IsArray() { return rawJSON } type functionRef struct { id string name string } edits := make([]antigravityContentEdit, 0) var pending []functionRef validOffsets := true contents.ForEach(func(contentIndex, content gjson.Result) bool { parts := content.Get("parts") if !parts.IsArray() { pending = nil return true } var calls, responses []functionRef var responseParts, otherParts []json.RawMessage partCount := 0 hasOtherPart := false parts.ForEach(func(_, part gjson.Result) bool { partCount++ switch { case part.Get("functionCall").Exists(): calls = append(calls, functionRef{id: part.Get("functionCall.id").String(), name: part.Get("functionCall.name").String()}) case part.Get("functionResponse").Exists(): responses = append(responses, functionRef{id: part.Get("functionResponse.id").String(), name: part.Get("functionResponse.name").String()}) responseParts = append(responseParts, json.RawMessage(part.Raw)) default: hasOtherPart = true otherParts = append(otherParts, json.RawMessage(part.Raw)) } return true }) if partCount == 0 { pending = nil return true } if len(calls) > 0 && len(responses) == 0 { pending = calls return true } if len(responses) == 0 { if hasOtherPart { pending = nil } return true } if len(calls) > 0 { pending = nil return true } var contentJSON []byte contentChanged := false if len(pending) > 0 && len(responses) > 0 { ordered := make([]json.RawMessage, 0, partCount) used := make([]bool, len(responses)) for _, call := range pending { for responseIndex, response := range responses { if used[responseIndex] { continue } if (call.id != "" && response.id == call.id) || (call.id == "" && call.name != "" && response.name == call.name) { used[responseIndex] = true ordered = append(ordered, responseParts[responseIndex]) break } } } for responseIndex := range responses { if !used[responseIndex] { ordered = append(ordered, responseParts[responseIndex]) } } if len(ordered) == len(responseParts) { ordered = append(ordered, otherParts...) encoded, errMarshal := json.Marshal(ordered) if errMarshal == nil && !bytes.Equal(encoded, []byte(parts.Raw)) { contentJSON = []byte(content.Raw) if updated, errSet := sjson.SetRawBytes(contentJSON, "parts", encoded); errSet == nil { contentJSON = updated contentChanged = true } } } } pending = nil if !hasOtherPart && content.Get("role").String() != "model" { if contentJSON == nil { contentJSON = []byte(content.Raw) } if updated, errSet := sjson.SetBytes(contentJSON, "role", "model"); errSet == nil { contentJSON = updated contentChanged = true } } if !contentChanged { return true } start := content.Index end := start + len(content.Raw) if start < 0 || end < start || end > len(rawJSON) || !bytes.Equal(rawJSON[start:end], []byte(content.Raw)) { validOffsets = false } edits = append(edits, antigravityContentEdit{ index: contentIndex.Int(), start: start, end: end, replacement: contentJSON, }) return true }) return applyAntigravityIndexedEdits(rawJSON, edits, validOffsets) } // applyAntigravityIndexedEdits splices collected JSON fragments into the original // request with one body copy. Applying SJSON once per field made large histories // scale with history size multiplied by the number of edits. func applyAntigravityIndexedEdits(rawJSON []byte, edits []antigravityContentEdit, validOffsets bool) []byte { if len(edits) == 0 { return rawJSON } if !validOffsets { return applyAntigravityContentEditsWithSJSON(rawJSON, edits) } finalSize := len(rawJSON) cursor := 0 for _, edit := range edits { if edit.start < cursor { return applyAntigravityContentEditsWithSJSON(rawJSON, edits) } finalSize += len(edit.replacement) - (edit.end - edit.start) if finalSize < 0 { return applyAntigravityContentEditsWithSJSON(rawJSON, edits) } cursor = edit.end } out := make([]byte, 0, finalSize) cursor = 0 for _, edit := range edits { out = append(out, rawJSON[cursor:edit.start]...) out = append(out, edit.replacement...) cursor = edit.end } return append(out, rawJSON[cursor:]...) } // applyAntigravityContentEditsWithSJSON preserves the legacy path semantics if // a GJSON result cannot be proven to point into the original request bytes. func applyAntigravityContentEditsWithSJSON(rawJSON []byte, edits []antigravityContentEdit) []byte { out := rawJSON for _, edit := range edits { path := edit.path if path == "" { path = fmt.Sprintf("request.contents.%d", edit.index) } if updated, errSet := sjson.SetRawBytes(out, path, edit.replacement); errSet == nil { out = updated } } return out } // repairAntigravityGeminiFunctionResponseNames copies missing or placeholder // functionResponse names from the matching functionCall. Edits are applied to // each part in isolation, then spliced into the request with one body copy. func repairAntigravityGeminiFunctionResponseNames(rawJSON []byte) []byte { contents := util.GetGJSONBytesNoCopy(rawJSON, "request.contents") if !contents.IsArray() { return rawJSON } callIDToName := make(map[string]string) contents.ForEach(func(_, content gjson.Result) bool { parts := content.Get("parts") if !parts.IsArray() { return true } parts.ForEach(func(_, part gjson.Result) bool { fc := part.Get("functionCall") if fc.Exists() { id := strings.TrimSpace(fc.Get("id").String()) name := strings.TrimSpace(fc.Get("name").String()) if id != "" && name != "" && name != "unknown" { callIDToName[id] = name } } return true }) return true }) if len(callIDToName) == 0 { return rawJSON } edits := make([]antigravityContentEdit, 0) validOffsets := true contents.ForEach(func(contentIdx, content gjson.Result) bool { parts := content.Get("parts") if !parts.IsArray() { return true } parts.ForEach(func(partIdx, part gjson.Result) bool { fr := part.Get("functionResponse") if !fr.Exists() { return true } id := strings.TrimSpace(fr.Get("id").String()) name := strings.TrimSpace(fr.Get("name").String()) if id == "" || (name != "" && name != "unknown") { return true } realName, ok := callIDToName[id] if !ok { return true } updatedPart, errSet := sjson.SetBytes([]byte(part.Raw), "functionResponse.name", realName) if errSet != nil { return true } start := part.Index end := start + len(part.Raw) if start < 0 || end < start || end > len(rawJSON) || !bytes.Equal(rawJSON[start:end], []byte(part.Raw)) { validOffsets = false } edits = append(edits, antigravityContentEdit{ index: contentIdx.Int(), path: fmt.Sprintf("request.contents.%d.parts.%d", contentIdx.Int(), partIdx.Int()), start: start, end: end, replacement: updatedPart, }) return true }) return true }) return applyAntigravityIndexedEdits(rawJSON, edits, validOffsets) } func validateAntigravityRequestSignatures(ctx context.Context, modelName string, from sdktranslator.Format, rawJSON []byte) ([]byte, error) { if from.String() != "claude" { return rawJSON, nil } before := countClaudeThinkingBlocks(rawJSON) if antigravityUsesReasoningReplayCache(modelName) { rawJSON = antigravityclaude.StripInvalidGeminiSignatureThinkingBlocks(rawJSON) logAntigravitySignatureStrip(before, countClaudeThinkingBlocks(rawJSON), "provider_cleanup", "empty_or_non_gemini_signature") return rawJSON, nil } // Claude models accept only Claude-format thinking signatures. rawJSON = antigravityclaude.StripEmptySignatureThinkingBlocks(rawJSON) logAntigravitySignatureStrip(before, countClaudeThinkingBlocks(rawJSON), "prefix_cleanup", "empty_or_non_claude_signature") if cache.SignatureCacheEnabled() { return rawJSON, nil } if !cache.SignatureBypassStrictMode() { // Non-strict bypass: let the translator handle invalid signatures // by dropping unsigned thinking blocks silently (no 400). return rawJSON, nil } before = countClaudeThinkingBlocks(rawJSON) rawJSON = antigravityclaude.StripInvalidBypassSignatureThinkingBlocks(rawJSON) logAntigravitySignatureStrip(before, countClaudeThinkingBlocks(rawJSON), "strict_bypass", "invalid_antigravity_claude_signature") return rawJSON, nil } func hasAntigravityClaudeTypedWebSearchTool(payload []byte) bool { tools := util.GetGJSONBytesNoCopy(payload, "tools") if !tools.IsArray() { return false } for _, tool := range tools.Array() { switch tool.Get("type").String() { case "web_search_20250305", "web_search_20260209": return true } } return false } func hasAntigravityGoogleSearchTool(payload []byte) bool { tools := util.GetGJSONBytesNoCopy(payload, "request.tools") if !tools.IsArray() { return false } for _, tool := range tools.Array() { if tool.Get("googleSearch").Exists() { return true } } return false } func shouldResolveAntigravityWebSearchGroundingURLs(from sdktranslator.Format, originalRequestRawJSON, requestRawJSON []byte) bool { return from.String() == "claude" && hasAntigravityClaudeTypedWebSearchTool(originalRequestRawJSON) && hasAntigravityGoogleSearchTool(requestRawJSON) } func (e *AntigravityExecutor) resolveWebSearchGroundingURLs(ctx context.Context, auth *cliproxyauth.Auth, from sdktranslator.Format, originalRequestRawJSON, requestRawJSON, responseRawJSON []byte) []byte { if !shouldResolveAntigravityWebSearchGroundingURLs(from, originalRequestRawJSON, requestRawJSON) { return responseRawJSON } return helps.ResolveAntigravityGroundingURLs(ctx, e.cfg, auth, responseRawJSON) } func countClaudeThinkingBlocks(rawJSON []byte) int { messages := util.GetGJSONBytesNoCopy(rawJSON, "messages") if !messages.IsArray() { return 0 } count := 0 messages.ForEach(func(_, message gjson.Result) bool { content := message.Get("content") if !content.IsArray() { return true } content.ForEach(func(_, part gjson.Result) bool { if part.Get("type").String() == "thinking" { count++ } return true }) return true }) return count } func logAntigravitySignatureStrip(before, after int, stage, reason string) { removed := before - after if removed <= 0 { return } log.WithFields(log.Fields{ "component": "signature_sanitizer", "executor": "antigravity", "target_provider": "claude", "action": "drop_thinking_blocks", "stage": stage, "reason": reason, "count": removed, }).Debug("antigravity executor: dropped Claude thinking blocks with invalid signatures") } // Identifier returns the executor identifier. func (e *AntigravityExecutor) Identifier() string { return antigravityAuthType } // PrepareRequest injects Antigravity credentials into the outgoing HTTP request. func (e *AntigravityExecutor) PrepareRequest(req *http.Request, auth *cliproxyauth.Auth) error { if req == nil { return nil } token, _, errToken := e.ensureAccessToken(req.Context(), auth) if errToken != nil { return errToken } if strings.TrimSpace(token) == "" { return statusErr{code: http.StatusUnauthorized, msg: "missing access token"} } req.Header.Set("Authorization", "Bearer "+token) return nil } // HttpRequest injects Antigravity credentials into the request and executes it. // It uses a whitelist approach: all incoming headers are stripped and only // the minimum set required by the Antigravity protocol is explicitly set. func (e *AntigravityExecutor) HttpRequest(ctx context.Context, auth *cliproxyauth.Auth, req *http.Request) (*http.Response, error) { if req == nil { return nil, fmt.Errorf("antigravity executor: request is nil") } if ctx == nil { ctx = req.Context() } httpReq := req.WithContext(ctx) // Connection management is a Request field, not a header, so the header // whitelist below cannot strip it. An inbound "Connection: close" makes Go's // server set Request.Close, and WithContext copies that field verbatim, which // would both leak the downstream header upstream and drain the shared pool. httpReq.Close = false // --- Whitelist: save only the headers we need from the original request --- contentType := httpReq.Header.Get("Content-Type") // Wipe ALL incoming headers for k := range httpReq.Header { delete(httpReq.Header, k) } // --- Set only the headers Antigravity actually sends --- if contentType != "" { httpReq.Header.Set("Content-Type", contentType) } // Content-Length is managed automatically by Go's http.Client from the Body httpReq.Header.Set("User-Agent", resolveUserAgent(auth)) // Inject Authorization: Bearer if err := e.PrepareRequest(httpReq, auth); err != nil { return nil, err } httpClient := newAntigravityHTTPClient(ctx, e.cfg, auth, 0) return httpClient.Do(httpReq) }