llm-cli-gateway 2.5.0 → 2.6.3

package/dist/index.js

@@ -7,7 +7,7 @@ import { existsSync, mkdirSync, readFileSync, readdirSync, renameSync, unlinkSyn
 import { dirname, join } from "path";
 import { fileURLToPath } from "url";
 import { z } from "zod/v3";
-import { executeCli, killAllProcessGroups } from "./executor.js";
+import { executeCli, killAllProcessGroups, providerCommandName } from "./executor.js";
 import { parseStreamJson } from "./stream-json-parser.js";
 import { parseCodexJsonStream } from "./codex-json-parser.js";
 import { parseGeminiJson, parseGeminiStreamJson } from "./gemini-json-parser.js";
@@ -27,7 +27,7 @@ import { createJobStore } from "./job-store.js";
 import { ApprovalManager } from "./approval-manager.js";
 import { checkReviewIntegrity } from "./review-integrity.js";
 import { buildClaudeMcpConfig, CLAUDE_MCP_SERVER_NAMES, } from "./claude-mcp-config.js";
-import { resolveGrokSessionArgs, resolveMistralSessionArgs, resolveCodexSessionArgs, sanitizeCliArgValues, prepareMistralRequest as buildMistralCliInvocation, MISTRAL_AGENT_MODES, GATEWAY_SESSION_PREFIX, resolveClaudePermissionFlags, resolveCodexSandboxFlags, CLAUDE_PERMISSION_MODES, GEMINI_APPROVAL_MODES, CODEX_SANDBOX_MODES, CODEX_ASK_FOR_APPROVAL_MODES, CLAUDE_EFFORT_LEVELS, prepareClaudeHighImpactFlags, validateClaudeAgentsMap, prepareCodexHighImpactFlags, prepareCodexForkRequest, CODEX_CONFIG_OVERRIDES_SCHEMA, prepareGeminiHighImpactFlags, prependGeminiAttachments, resolveGeminiSessionPlan, GEMINI_HIGH_IMPACT_PARAMS_SCHEMA, } from "./request-helpers.js";
+import { resolveGrokSessionArgs, resolveMistralSessionArgs, resolveCodexSessionArgs, sanitizeCliArgValues, prepareMistralRequest as buildMistralCliInvocation, MISTRAL_AGENT_MODES, GATEWAY_SESSION_PREFIX, resolveClaudePermissionFlags, resolveCodexSandboxFlags, CLAUDE_PERMISSION_MODES, GEMINI_APPROVAL_MODES, CODEX_SANDBOX_MODES, CODEX_ASK_FOR_APPROVAL_MODES, CLAUDE_EFFORT_LEVELS, prepareClaudeHighImpactFlags, validateClaudeAgentsMap, prepareCodexHighImpactFlags, prepareCodexForkRequest, CODEX_CONFIG_OVERRIDES_SCHEMA, resolveGeminiSessionPlan, GEMINI_HIGH_IMPACT_PARAMS_SCHEMA, } from "./request-helpers.js";
 import { createFlightRecorder } from "./flight-recorder.js";
 import { resolvePromptInput, PromptPartsSchema, assembleClaudeCacheBlocks, } from "./prompt-parts.js";
 import { computeSessionCacheStats, computeTtlRemaining, readPersistedRequest, PERSISTED_REQUEST_DEFAULT_MAX_CHARS, } from "./cache-stats.js";
@@ -38,7 +38,8 @@ import { printDoctorJson } from "./doctor.js";
 import { createWorkspace, describeWorkspace, getWorkspace, loadWorkspaceRegistry, registerExistingWorkspace, resolveWorkspaceForProvider, validatePathInsideWorkspace, } from "./workspace-registry.js";
 import { generateSecret, hashSecret } from "./oauth.js";
 import { registerValidationTools } from "./validation-tools.js";
-import { assertUpstreamCliArgs, assertUpstreamCliEnv, buildProviderSubcommandsCompactCatalog, buildUpstreamContractReport, getCliSubcommandContract, probeInstalledCliContract, serializeCliSubcommandContract, } from "./upstream-contracts.js";
+import { assertUpstreamCliArgs, assertUpstreamCliEnv, buildProviderSubcommandsCompactCatalog, buildUpstreamContractReport, getCliSubcommandContract, probeInstalledCliContract, serializeCliSubcommandContract, UPSTREAM_CLI_CONTRACTS, } from "./upstream-contracts.js";
+import { buildArgvFromGeneration, deriveZodShapeFromGeneration, GROK_FLAG_GENERATION, GROK_GEN_OUTPUT_FORMAT, GROK_GEN_MAIN, GROK_GEN_PROMPT_FILE, GROK_GEN_SINGLE, GROK_GEN_TAIL, } from "./provider-codegen.js";
 import { entrypointFileURL } from "./entrypoint-url.js";
 const logger = {
     info: (message, ...args) => {
@@ -162,7 +163,7 @@ Other: list_models, cli_versions, upstream_contracts, provider_subcommands_* (re
 
 Key behaviors:
 ${deferralLine}
-- Sessions: Claude --continue, Gemini --resume, Grok --resume/--continue, Mistral --resume/--continue (current Vibe defaults session logging on; doctor flags explicit session_logging.enabled=false), Codex \`exec resume <ID>\` / \`exec resume --last\` (all real CLI continuity). For Codex, sessionId must be a real Codex UUID (from ~/.codex/sessions/); gateway-generated gw-* IDs are rejected.
+- Sessions: Claude --continue, Gemini (Antigravity) --conversation <id>/--continue, Grok --resume/--continue, Mistral --resume/--continue (current Vibe defaults session logging on; doctor flags explicit session_logging.enabled=false), Codex \`exec resume <ID>\` / \`exec resume --last\` (all real CLI continuity). For Codex, sessionId must be a real Codex UUID (from ~/.codex/sessions/); gateway-generated gw-* IDs are rejected.
 - Approval gates: opt-in via approvalStrategy:"mcp_managed".
 - Upstream drift detection: After upgrading any provider CLI (especially grok), use upstream_contracts with probeInstalled:true and provider_subcommand_drift for declared subcommand help surfaces. Probes are safe, read-only --help checks.
 - Idle timeout kills stuck processes (default 10min, configurable via idleTimeoutMs).
@@ -230,6 +231,7 @@ function getApprovalManager(runtimeLogger = logger) {
 const MCP_SERVER_ENUM = z.enum(CLAUDE_MCP_SERVER_NAMES);
 const CLI_TYPE_ENUM = z.enum(CLI_TYPES);
 export const MAX_TURNS_SCHEMA = z.number().int().positive().safe().max(10_000);
+const GROK_GENERATED_SHAPE = deriveZodShapeFromGeneration(UPSTREAM_CLI_CONTRACTS.grok, GROK_FLAG_GENERATION);
 export const MAX_TOKENS_SCHEMA = z.number().int().positive().safe().max(100_000_000);
 export const MAX_PRICE_SCHEMA = z.number().positive().finite().min(1e-6).max(10_000);
 export const WORKTREE_SCHEMA = z
@@ -340,7 +342,7 @@ async function awaitJobOrDefer(cli, args, corrId, idleTimeoutMs, outputFormat, f
         runtime.persistence.asyncJobsEnabled &&
         runtime.asyncJobManager.hasStore();
     if (SYNC_DEADLINE_MS === 0 || !deferralAvailable) {
-        const command = cli === "mistral" ? "vibe" : cli;
+        const command = providerCommandName(cli);
         try {
             return await executeCli(command, args, {
                 idleTimeout: idleTimeoutMs,
@@ -687,6 +689,7 @@ function createErrorResponse(cli, code, stderr, correlationId, error) {
         content: [{ type: "text", text: errorMessage }],
         isError: true,
         structuredContent: {
+            response: errorMessage,
             correlationId: correlationId || null,
             cli,
             exitCode: code,
@@ -758,6 +761,7 @@ function buildAsyncFlightRecorderHandoff(cliName, prep, sessionId, outputFormat)
             stablePrefixHash: prep.stablePrefixHash ?? undefined,
             stablePrefixTokens: prep.stablePrefixTokens ?? undefined,
             cacheControlBlocks: prep.cacheControlBlocks,
+            cacheControlTtlSeconds: prep.cacheControlTtlSeconds,
         },
         extractUsage: (stdout) => extractUsageAndCost(cli, stdout, fmt, { sessionId: sid, home }),
     };
@@ -1102,7 +1106,18 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
     const ccEarly = params.promptParts?.cacheControl;
     const cacheControlRequestedEarly = !!(ccEarly &&
         (ccEarly.system || ccEarly.tools || ccEarly.context));
-    if (params.optimizePrompt && cacheControlRequestedEarly) {
+    const explicitCacheControlBlockCount = params.promptParts && ccEarly
+        ? (ccEarly.system && params.promptParts.system && params.promptParts.system.length > 0
+            ? 1
+            : 0) +
+            (ccEarly.tools && params.promptParts.tools && params.promptParts.tools.length > 0 ? 1 : 0) +
+            (ccEarly.context && params.promptParts.context && params.promptParts.context.length > 0
+                ? 1
+                : 0)
+        : 0;
+    const effectiveExplicitCacheControl = explicitCacheControlBlockCount > 0;
+    const cacheControlNoop = cacheControlRequestedEarly && !effectiveExplicitCacheControl;
+    if (params.optimizePrompt && effectiveExplicitCacheControl) {
         return createErrorResponse(params.operation, 1, "", corrId, new Error("optimizePrompt is incompatible with promptParts.cacheControl (slice κ): optimization rewrites the assembled prompt text the flight recorder logs, while the cache_control payload is built from raw promptParts; the two would desync and break Anthropic prefix-cache reuse. Disable optimizePrompt when opting into cacheControl."));
     }
     let effectivePrompt = assembledPrompt;
@@ -1111,7 +1126,7 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
         logOptimizationTokens("prompt", corrId, effectivePrompt, optimized);
         effectivePrompt = optimized;
     }
-    const requestedMcpServers = normalizeMcpServers(params.mcpServers);
+    const requestedMcpServers = params.mcpServers ? [...new Set(params.mcpServers)] : [];
     const mcpConfigResolution = resolveClaudeMcpConfig(params.operation, corrId, requestedMcpServers, params.strictMcpConfig);
     if ("errorResponse" in mcpConfigResolution) {
         return mcpConfigResolution.errorResponse;
@@ -1137,7 +1152,7 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
         }
     }
     let autoEmittedCacheControlBlock = null;
-    if (!cacheControlRequestedEarly &&
+    if (!effectiveExplicitCacheControl &&
         runtime.cacheAwareness.emitAnthropicCacheControl &&
         !params.optimizePrompt &&
         params.outputFormat === "stream-json" &&
@@ -1161,7 +1176,14 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
         }
     }
     const warnings = [];
-    if (!cacheControlRequestedEarly &&
+    if (cacheControlNoop) {
+        warnings.push({
+            code: "cache_control_noop",
+            message: "promptParts.cacheControl only marked empty or omitted stable parts; no cache_control breakpoint will be emitted from the explicit marker.",
+            reason: "cacheControl marker did not match a non-empty stable block",
+        });
+    }
+    if (!effectiveExplicitCacheControl &&
         autoEmittedCacheControlBlock === null &&
         params.promptParts &&
         stablePrefixTokens !== null) {
@@ -1181,9 +1203,10 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
             });
         }
     }
-    const cacheControlRequested = cacheControlRequestedEarly || autoEmittedCacheControlBlock !== null;
+    const cacheControlRequested = effectiveExplicitCacheControl || autoEmittedCacheControlBlock !== null;
     let stdinPayload;
     let cacheControlBlocks;
+    let cacheControlTtlSeconds;
     if (cacheControlRequested) {
         if (params.outputFormat !== "stream-json") {
             return createErrorResponse(params.operation, 1, "", corrId, new Error("promptParts.cacheControl requires outputFormat: 'stream-json' (slice κ pipes the cache_control blocks over --input-format stream-json; text/json output formats cannot carry the required NDJSON usage events)."));
@@ -1200,6 +1223,7 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
         const built = assembleClaudeCacheBlocks(effectiveParts);
         stdinPayload = `${JSON.stringify(built.payload)}\n`;
         cacheControlBlocks = built.markedBlockCount;
+        cacheControlTtlSeconds = built.markedBlockCount > 0 ? 3600 : undefined;
     }
     const args = cacheControlRequested
         ? [
@@ -1288,6 +1312,7 @@ export function prepareClaudeRequest(params, runtime = resolveGatewayServerRunti
         stablePrefixTokens,
         stdinPayload,
         cacheControlBlocks,
+        cacheControlTtlSeconds,
         warnings: warnings.length > 0 ? warnings : undefined,
     };
 }
@@ -1320,7 +1345,7 @@ export function prepareCodexRequest(params, runtime = resolveGatewayServerRuntim
         logOptimizationTokens("prompt", corrId, effectivePrompt, optimized);
         effectivePrompt = optimized;
     }
-    const requestedMcpServers = normalizeMcpServers(params.mcpServers);
+    const requestedMcpServers = params.mcpServers ? [...new Set(params.mcpServers)] : [];
     let approvalDecision = null;
     if (params.approvalStrategy === "mcp_managed") {
         approvalDecision = runtime.approvalManager.decide({
@@ -1478,7 +1503,7 @@ export function prepareGeminiRequest(params, runtime = resolveGatewayServerRunti
         logOptimizationTokens("prompt", corrId, effectivePrompt, optimized);
         effectivePrompt = optimized;
     }
-    const requestedMcpServers = normalizeMcpServers(params.mcpServers);
+    const requestedMcpServers = params.mcpServers ? [...new Set(params.mcpServers)] : [];
     let approvalDecision = null;
     if (params.approvalStrategy === "mcp_managed") {
         approvalDecision = runtime.approvalManager.decide({
@@ -1498,51 +1523,45 @@ export function prepareGeminiRequest(params, runtime = resolveGatewayServerRunti
         }
     }
     const effectiveApprovalMode = params.approvalStrategy === "mcp_managed" ? "yolo" : params.approvalMode;
-    const highImpact = prepareGeminiHighImpactFlags({
-        sandbox: params.sandbox,
-        policyFiles: params.policyFiles,
-        adminPolicyFiles: params.adminPolicyFiles,
-    });
-    if (highImpact.missingPolicyPath) {
-        return createErrorResponse(params.operation, 1, "", corrId, new Error(`${highImpact.missingPolicyField}: path does not exist: ${highImpact.missingPolicyPath}`));
-    }
-    if (params.attachments && params.attachments.length > 0) {
-        try {
-            effectivePrompt = prependGeminiAttachments(effectivePrompt, params.attachments);
-        }
-        catch (err) {
-            return createErrorResponse(params.operation, 1, "", corrId, err instanceof Error ? err : new Error(String(err)));
-        }
-    }
-    const args = ["-p", effectivePrompt];
-    if (resolvedModel)
-        args.push("--model", resolvedModel);
-    if (effectiveApprovalMode)
-        args.push("--approval-mode", effectiveApprovalMode);
-    if (params.yolo && effectiveApprovalMode !== "yolo") {
-        args.push("--yolo");
+    const unsupported = (field, detail) => createErrorResponse(params.operation, 1, "", corrId, new Error(`${field} is not supported by Antigravity CLI (agy): ${detail}`));
+    if (effectiveApprovalMode &&
+        effectiveApprovalMode !== "default" &&
+        effectiveApprovalMode !== "yolo") {
+        return unsupported("approvalMode", "use 'default' for prompted execution or 'yolo'/yolo=true for --dangerously-skip-permissions");
     }
     if (params.allowedTools && params.allowedTools.length > 0) {
-        sanitizeCliArgValues(params.allowedTools, "allowedTools");
-        params.allowedTools.forEach(tool => args.push("--allowed-tools", tool));
+        return unsupported("allowedTools", "agy has no non-interactive allowed-tools flag");
     }
     if (requestedMcpServers.length > 0) {
-        sanitizeCliArgValues(requestedMcpServers, "mcpServers");
-        requestedMcpServers.forEach(serverName => args.push("--allowed-mcp-server-names", serverName));
+        return unsupported("mcpServers", "agy has no non-interactive allowed MCP server allowlist flag");
     }
-    if (params.includeDirs && params.includeDirs.length > 0) {
-        sanitizeCliArgValues(params.includeDirs, "includeDirs");
-        params.includeDirs.forEach(dir => args.push("--include-directories", dir));
+    if (params.outputFormat && params.outputFormat !== "text") {
+        return unsupported("outputFormat", "agy print mode currently emits text only");
     }
-    args.push(...highImpact.args);
-    if (params.outputFormat === "json") {
-        args.push("-o", "json");
+    if (params.policyFiles && params.policyFiles.length > 0) {
+        return unsupported("policyFiles", "agy has no --policy flag");
     }
-    else if (params.outputFormat === "stream-json") {
-        args.push("-o", "stream-json");
+    if (params.adminPolicyFiles && params.adminPolicyFiles.length > 0) {
+        return unsupported("adminPolicyFiles", "agy has no --admin-policy flag");
+    }
+    if (params.attachments && params.attachments.length > 0) {
+        return unsupported("attachments", "agy has no documented @path attachment-token contract");
     }
     if (params.skipTrust) {
-        args.push("--skip-trust");
+        return unsupported("skipTrust", "agy has no --skip-trust flag");
+    }
+    const args = ["--print", effectivePrompt];
+    if (resolvedModel)
+        args.push("--model", resolvedModel);
+    if (params.includeDirs && params.includeDirs.length > 0) {
+        sanitizeCliArgValues(params.includeDirs, "includeDirs");
+        params.includeDirs.forEach(dir => args.push("--add-dir", dir));
+    }
+    if (params.sandbox) {
+        args.push("--sandbox");
+    }
+    if (params.yolo || effectiveApprovalMode === "yolo") {
+        args.push("--dangerously-skip-permissions");
     }
     return {
         corrId,
@@ -1611,76 +1630,19 @@ export function prepareGrokRequest(params, runtime = resolveGatewayServerRuntime
         }
     }
     const effectiveAlwaysApprove = params.approvalStrategy === "mcp_managed" ? true : Boolean(params.alwaysApprove);
+    const grokContract = UPSTREAM_CLI_CONTRACTS.grok;
+    const genParams = params;
     const args = ["-p", effectivePrompt];
     if (resolvedModel)
         args.push("--model", resolvedModel);
-    if (params.outputFormat)
-        args.push("--output-format", params.outputFormat);
+    args.push(...buildArgvFromGeneration(grokContract, GROK_GEN_OUTPUT_FORMAT, genParams));
     if (effectiveAlwaysApprove) {
         args.push("--always-approve");
     }
     else if (params.permissionMode) {
         args.push("--permission-mode", params.permissionMode);
     }
-    if (params.effort)
-        args.push("--effort", params.effort);
-    if (params.reasoningEffort)
-        args.push("--reasoning-effort", params.reasoningEffort);
-    if (params.allowedTools && params.allowedTools.length > 0) {
-        args.push("--tools", params.allowedTools.join(","));
-    }
-    if (params.disallowedTools && params.disallowedTools.length > 0) {
-        args.push("--disallowed-tools", params.disallowedTools.join(","));
-    }
-    if (params.maxTurns !== undefined) {
-        args.push("--max-turns", String(params.maxTurns));
-    }
-    if (params.workingDir) {
-        args.push("--cwd", params.workingDir);
-    }
-    if (params.sandbox) {
-        args.push("--sandbox", params.sandbox);
-    }
-    if (params.rules) {
-        args.push("--rules", params.rules);
-    }
-    if (params.systemPromptOverride) {
-        args.push("--system-prompt-override", params.systemPromptOverride);
-    }
-    if (params.allow && params.allow.length > 0) {
-        for (const rule of params.allow) {
-            args.push("--allow", rule);
-        }
-    }
-    if (params.deny && params.deny.length > 0) {
-        for (const rule of params.deny) {
-            args.push("--deny", rule);
-        }
-    }
-    if (params.compactionMode) {
-        args.push("--compaction-mode", params.compactionMode);
-    }
-    if (params.compactionDetail) {
-        args.push("--compaction-detail", params.compactionDetail);
-    }
-    if (params.agent) {
-        args.push("--agent", params.agent);
-    }
-    if (params.bestOfN !== undefined) {
-        args.push("--best-of-n", String(params.bestOfN));
-    }
-    if (params.check) {
-        args.push("--check");
-    }
-    if (params.disableWebSearch) {
-        args.push("--disable-web-search");
-    }
-    if (params.todoGate) {
-        args.push("--todo-gate");
-    }
-    if (params.verbatim) {
-        args.push("--verbatim");
-    }
+    args.push(...buildArgvFromGeneration(grokContract, GROK_GEN_MAIN, genParams));
     if (params.agents !== undefined) {
         if (typeof params.agents === "string") {
             if (!params.agents.trim()) {
@@ -1696,9 +1658,7 @@ export function prepareGrokRequest(params, runtime = resolveGatewayServerRuntime
             args.push("--agents", JSON.stringify(agentsResult.value));
         }
     }
-    if (params.promptFile) {
-        args.push("--prompt-file", params.promptFile);
-    }
+    args.push(...buildArgvFromGeneration(grokContract, GROK_GEN_PROMPT_FILE, genParams));
     if (params.promptJson !== undefined) {
         const promptJsonValue = typeof params.promptJson === "string" ? params.promptJson : JSON.stringify(params.promptJson);
         if (!promptJsonValue.trim()) {
@@ -1706,33 +1666,8 @@ export function prepareGrokRequest(params, runtime = resolveGatewayServerRuntime
         }
         args.push("--prompt-json", promptJsonValue);
     }
-    if (params.single) {
-        args.push("--single", params.single);
-    }
-    if (params.experimentalMemory) {
-        args.push("--experimental-memory");
-    }
-    if (params.noAltScreen) {
-        args.push("--no-alt-screen");
-    }
-    if (params.noMemory) {
-        args.push("--no-memory");
-    }
-    if (params.noPlan) {
-        args.push("--no-plan");
-    }
-    if (params.noSubagents) {
-        args.push("--no-subagents");
-    }
-    if (params.oauth) {
-        args.push("--oauth");
-    }
-    if (params.restoreCode) {
-        args.push("--restore-code");
-    }
-    if (params.leaderSocket) {
-        args.push("--leader-socket", params.leaderSocket);
-    }
+    args.push(...buildArgvFromGeneration(grokContract, GROK_GEN_SINGLE, genParams));
+    args.push(...buildArgvFromGeneration(grokContract, GROK_GEN_TAIL, genParams));
     if (params.nativeWorktree === true) {
         args.push("--worktree");
     }
@@ -1889,6 +1824,7 @@ function buildCliResponse(cli, stdout, optimizeResponse, corrId, sessionId, prep
     const response = {
         content: [{ type: "text", text: finalStdout }],
         structuredContent: {
+            response: finalStdout,
             model: prep.resolvedModel || "default",
             cli,
             correlationId: corrId,
@@ -2024,6 +1960,7 @@ function buildGrokApiToolResponse(args) {
     const response = {
         content: [{ type: "text", text }],
         structuredContent: {
+            response: text,
             provider: "grok-api",
             cli: "grok-api",
             model: args.result.model || args.prep.resolvedModel,
@@ -3468,6 +3405,7 @@ export function createGatewayServer(deps = {}) {
             stablePrefixHash: prep.stablePrefixHash ?? undefined,
             stablePrefixTokens: prep.stablePrefixTokens ?? undefined,
             cacheControlBlocks: prep.cacheControlBlocks,
+            cacheControlTtlSeconds: prep.cacheControlTtlSeconds,
         }, runtime);
         logger.info(`[${corrId}] claude_request invoked with model=${prep.resolvedModel || "default"}, outputFormat=${outputFormat}, prompt length=${prep.effectivePrompt.length}, sessionId=${effectiveSessionId}, cacheControlBlocks=${prep.cacheControlBlocks ?? 0}`);
         try {
@@ -3967,23 +3905,23 @@ export function createGatewayServer(deps = {}) {
             performanceMetrics.recordRequest("codex", finalizedDurationMs, wasSuccessful);
         }
     });
-    server.tool("gemini_request", "Run a Google Gemini CLI request synchronously (when async jobs are enabled, auto-defers to a pollable job past the sync deadline; otherwise runs to completion). Requires exactly one of prompt or promptParts.", {
+    server.tool("gemini_request", "Run a Google Antigravity CLI (`agy`) request through the Gemini-compatible gateway tool synchronously (when async jobs are enabled, auto-defers to a pollable job past the sync deadline; otherwise runs to completion). Requires exactly one of prompt or promptParts.", {
         prompt: z
             .string()
             .min(1, "Prompt cannot be empty")
             .max(100000, "Prompt too long (max 100k chars)")
             .optional()
-            .describe("Prompt text for Gemini (mutually exclusive with promptParts)"),
+            .describe("Prompt text for Antigravity CLI (mutually exclusive with promptParts)"),
         promptParts: PromptPartsSchema.optional().describe("Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking."),
         model: z
             .string()
             .optional()
-            .describe("Model name or alias (e.g. gemini-3-pro-preview, gemini-2.5-flash, pro, flash, latest)"),
+            .describe("Model name or alias passed to agy --model (e.g. gemini-3-pro-preview, gemini-2.5-flash, pro, flash, latest)"),
         sessionId: z
             .string()
             .optional()
-            .describe("Gemini session ID to resume (emits --resume <id>), or 'latest' for the most recent session in this cwd"),
-        resumeLatest: z.boolean().default(false).describe("Resume latest session"),
+            .describe("Antigravity conversation ID to resume (emits --conversation <id>)"),
+        resumeLatest: z.boolean().default(false).describe("Continue the most recent conversation"),
         createNewSession: z.boolean().default(false).describe("Force new session"),
         approvalMode: z
             .enum(GEMINI_APPROVAL_MODES)
@@ -3999,13 +3937,16 @@ export function createGatewayServer(deps = {}) {
             .describe("Approval policy override"),
         mcpServers: z
             .array(MCP_SERVER_ENUM)
-            .default(["sqry"])
-            .describe("MCP server names passed to Gemini as --allowed-mcp-server-names"),
+            .default([])
+            .describe("Unsupported for Antigravity CLI; non-empty values are rejected"),
         allowedTools: z
             .array(z.string())
             .optional()
-            .describe("Allowed tools (['Write','Edit','Bash'])"),
-        includeDirs: z.array(z.string()).optional().describe("Additional workspace directories"),
+            .describe("Unsupported for Antigravity CLI; non-empty values are rejected"),
+        includeDirs: z
+            .array(z.string())
+            .optional()
+            .describe("Additional workspace directories passed as --add-dir"),
         correlationId: z.string().optional().describe("Request trace ID (auto if omitted)"),
         optimizePrompt: z.boolean().default(false).describe("Optimize prompt before execution"),
         optimizeResponse: z.boolean().default(false).describe("Optimize response output"),
@@ -4023,19 +3964,19 @@ export function createGatewayServer(deps = {}) {
         outputFormat: z
             .enum(["text", "json", "stream-json"])
             .default("text")
-            .describe("Gemini output format. `json` emits `-o json` (single JSON with usageMetadata). `stream-json` emits `-o stream-json` (NDJSON event stream — `init`/`message`/`result` lines, usage extracted from the terminal `result.stats` event). Both report usage to the flight recorder."),
-        sandbox: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.sandbox.describe("Run Gemini in sandbox mode (-s)"),
-        policyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.policyFiles.describe("Policy file paths (--policy <path>, one per file). Paths must exist."),
-        adminPolicyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.adminPolicyFiles.describe("Admin policy file paths (--admin-policy <path>, one per file). Paths must exist."),
-        attachments: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.attachments.describe("Absolute file paths prepended as @<path> tokens to the prompt"),
+            .describe("Antigravity CLI currently supports text output only through the gateway; json and stream-json are rejected."),
+        sandbox: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.sandbox.describe("Run Antigravity in sandbox mode (--sandbox)"),
+        policyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.policyFiles.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
+        adminPolicyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.adminPolicyFiles.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
+        attachments: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.attachments.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
         skipTrust: z
             .boolean()
             .default(false)
-            .describe("Emit `--skip-trust` so Gemini trusts the workspace for this session and skips the interactive trust prompt (Phase 4 slice γ). Required for headless runs in fresh workspaces."),
+            .describe("Unsupported for Antigravity CLI; true is rejected."),
         yolo: z
             .boolean()
             .optional()
-            .describe("Emit `--yolo` to auto-approve all actions. Equivalent to approvalMode 'yolo'; routed through the same approval gate. Under mcp_managed the gate still decides."),
+            .describe("Emit `--dangerously-skip-permissions` to auto-approve all actions. Routed through the same approval gate. Under mcp_managed the gate still decides."),
         workspace: WORKSPACE_ALIAS_SCHEMA.optional(),
         worktree: WORKTREE_SCHEMA.optional(),
     }, {
@@ -4083,10 +4024,7 @@ export function createGatewayServer(deps = {}) {
             .describe("Prompt text for Grok (mutually exclusive with promptParts)"),
         promptParts: PromptPartsSchema.optional().describe("Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking."),
         model: z.string().optional().describe("Model name or alias (e.g. grok-build, latest)"),
-        outputFormat: z
-            .enum(["plain", "json", "streaming-json"])
-            .optional()
-            .describe("Output format (plain|json|streaming-json). Grok default is plain."),
+        ...GROK_GENERATED_SHAPE,
         sessionId: z
             .string()
             .optional()
@@ -4104,11 +4042,6 @@ export function createGatewayServer(deps = {}) {
             .enum(["default", "acceptEdits", "auto", "dontAsk", "bypassPermissions", "plan"])
             .optional()
             .describe("Grok permission mode"),
-        effort: z
-            .enum(["low", "medium", "high", "xhigh", "max"])
-            .optional()
-            .describe("Grok effort level"),
-        reasoningEffort: z.string().optional().describe("Reasoning effort for reasoning models"),
         approvalStrategy: z
             .enum(["legacy", "mcp_managed"])
             .default("legacy")
@@ -4121,14 +4054,6 @@ export function createGatewayServer(deps = {}) {
             .array(MCP_SERVER_ENUM)
             .default(["sqry"])
             .describe("MCP server names for approval tracking (Grok manages its own MCP config via `grok mcp`)"),
-        allowedTools: z
-            .array(z.string())
-            .optional()
-            .describe("Allowed built-in tools (passed as --tools comma list)"),
-        disallowedTools: z
-            .array(z.string())
-            .optional()
-            .describe("Disallowed built-in tools (passed as --disallowed-tools comma list)"),
         correlationId: z.string().optional().describe("Request trace ID (auto if omitted)"),
         optimizePrompt: z.boolean().default(false).describe("Optimize prompt before execution"),
         optimizeResponse: z.boolean().default(false).describe("Optimize response output"),
@@ -4143,107 +4068,14 @@ export function createGatewayServer(deps = {}) {
             .boolean()
             .default(false)
             .describe("Bypass dedup and force a fresh CLI run even if a recent identical request exists"),
-        maxTurns: MAX_TURNS_SCHEMA.optional().describe("Grok `--max-turns N`: cap on agent-loop iterations for cost / latency control (Phase 4 slice δ). Bounded to safe integers ≤ 10000."),
-        workingDir: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --cwd <DIR>: working directory for this invocation. Lets headless callers run Grok against a directory other than the gateway process's cwd."),
-        sandbox: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --sandbox <PROFILE>: sandbox profile for filesystem and network access. Freeform per `grok --help` (no enum constraint on Grok 0.1.210); also settable via GROK_SANDBOX env var. Caller responsibility to pass a valid profile name."),
-        rules: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --rules <RULES>: extra rules to append to the system prompt. Supports `@file` prefix per `grok --help` to load from a file; gateway passes the value verbatim and lets Grok parse the prefix."),
-        systemPromptOverride: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --system-prompt-override <PROMPT>: replace the agent's system prompt entirely. Distinct from Claude's --system-prompt / --append-system-prompt (Grok has only one override flag, not a pair)."),
-        allow: z
-            .array(z.string())
-            .optional()
-            .describe('Grok --allow <RULE>: permission allow rules. Each entry is emitted as its own --allow instance (per `grok --help`: "Repeat to add multiple rules").'),
-        deny: z
-            .array(z.string())
-            .optional()
-            .describe('Grok --deny <RULE>: permission deny rules. Each entry is emitted as its own --deny instance (per `grok --help`: "Repeat to add multiple rules").'),
-        compactionMode: z
-            .enum(["summary", "transcript", "segments"])
-            .optional()
-            .describe("Grok --compaction-mode: summary (default; no pointer) | transcript (points at the raw transcript) | segments (persists per-segment markdown to grep). Sets GROK_COMPACTION_MODE."),
-        compactionDetail: z
-            .enum(["none", "minimal", "balanced", "verbose"])
-            .optional()
-            .describe("Grok --compaction-detail: verbatim segment detail (none|minimal|balanced|verbose, default verbose). Only affects `--compaction-mode segments`. Sets GROK_COMPACTION_DETAIL."),
-        agent: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --agent <NAME>: agent name or definition file path."),
-        bestOfN: MAX_TURNS_SCHEMA.optional().describe("Grok --best-of-n <N>: run the task N ways in parallel and pick the best (headless only)."),
-        check: z
-            .boolean()
-            .optional()
-            .describe("Grok --check: append a self-verification loop to the prompt (headless only)."),
-        disableWebSearch: z
-            .boolean()
-            .optional()
-            .describe("Grok --disable-web-search: disable web search and remote retrieval tools."),
-        todoGate: z
-            .boolean()
-            .optional()
-            .describe("Grok --todo-gate: enable runtime turn-end TodoGate for this session (session-scoped, not persisted)."),
-        verbatim: z
-            .boolean()
-            .optional()
-            .describe("Grok --verbatim: send the prompt exactly as given. Also skips gateway optimizePrompt when true."),
         agents: z
             .union([z.string().min(1), z.record(z.string(), z.record(z.string(), z.unknown()))])
             .optional()
             .describe("Grok --agents <JSON>: inline subagent definitions (JSON string or name → { description, prompt, … } map)."),
-        promptFile: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --prompt-file <PATH>: single-turn prompt loaded from a file."),
         promptJson: z
             .union([z.string(), z.array(z.unknown()), z.record(z.string(), z.unknown())])
             .optional()
             .describe("Grok --prompt-json <JSON>: single-turn prompt JSON blocks (string or serializable value)."),
-        single: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok --single <PROMPT>: single-turn prompt (in addition to gateway -p)."),
-        experimentalMemory: z
-            .boolean()
-            .optional()
-            .describe("Grok --experimental-memory: enable cross-session memory."),
-        noAltScreen: z
-            .boolean()
-            .optional()
-            .describe("Grok --no-alt-screen: run inline without alt screen."),
-        noMemory: z.boolean().optional().describe("Grok --no-memory: disable cross-session memory."),
-        noPlan: z.boolean().optional().describe("Grok --no-plan: disable plan mode."),
-        noSubagents: z
-            .boolean()
-            .optional()
-            .describe("Grok --no-subagents: disable subagent spawning."),
-        oauth: z.boolean().optional().describe("Grok --oauth: use OAuth during authentication."),
-        restoreCode: z
-            .boolean()
-            .optional()
-            .describe("Grok --restore-code: check out the original session commit when resuming."),
-        leaderSocket: z
-            .string()
-            .min(1)
-            .optional()
-            .describe("Grok 0.2.32+ --leader-socket <PATH>: custom leader socket path (default ~/.grok/leader.sock). Targets an isolated leader process, e.g. a local/branch Grok build; name it ~/.grok/leader-*.sock to keep `grok leader list/kill` discovery working."),
         nativeWorktree: z
             .union([z.boolean(), z.string().min(1)])
             .optional()
@@ -4835,23 +4667,23 @@ export function createGatewayServer(deps = {}) {
                 worktree,
             });
         });
-        server.tool("gemini_request_async", "Start a Google Gemini CLI request as a durable background job. Poll with llm_job_status, collect with llm_job_result.", {
+        server.tool("gemini_request_async", "Start a Google Antigravity CLI (`agy`) request as a durable background job through the Gemini-compatible gateway tool. Poll with llm_job_status, collect with llm_job_result.", {
             prompt: z
                 .string()
                 .min(1, "Prompt cannot be empty")
                 .max(100000, "Prompt too long (max 100k chars)")
                 .optional()
-                .describe("Prompt text for Gemini (mutually exclusive with promptParts)"),
+                .describe("Prompt text for Antigravity CLI (mutually exclusive with promptParts)"),
             promptParts: PromptPartsSchema.optional().describe("Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking."),
             model: z
                 .string()
                 .optional()
-                .describe("Model name or alias (e.g. gemini-3-pro-preview, gemini-2.5-flash, pro, flash, latest)"),
+                .describe("Model name or alias passed to agy --model (e.g. gemini-3-pro-preview, gemini-2.5-flash, pro, flash, latest)"),
             sessionId: z
                 .string()
                 .optional()
-                .describe("Gemini session ID to resume (emits --resume <id>), or 'latest' for the most recent session in this cwd"),
-            resumeLatest: z.boolean().default(false).describe("Resume latest session"),
+                .describe("Antigravity conversation ID to resume (emits --conversation <id>)"),
+            resumeLatest: z.boolean().default(false).describe("Continue the most recent conversation"),
             createNewSession: z.boolean().default(false).describe("Force new session"),
             approvalMode: z
                 .enum(GEMINI_APPROVAL_MODES)
@@ -4867,13 +4699,16 @@ export function createGatewayServer(deps = {}) {
                 .describe("Approval policy override"),
             mcpServers: z
                 .array(MCP_SERVER_ENUM)
-                .default(["sqry"])
-                .describe("MCP server names passed to Gemini as --allowed-mcp-server-names"),
+                .default([])
+                .describe("Unsupported for Antigravity CLI; non-empty values are rejected"),
             allowedTools: z
                 .array(z.string())
                 .optional()
-                .describe("Allowed tools (['Write','Edit','Bash'])"),
-            includeDirs: z.array(z.string()).optional().describe("Additional workspace directories"),
+                .describe("Unsupported for Antigravity CLI; non-empty values are rejected"),
+            includeDirs: z
+                .array(z.string())
+                .optional()
+                .describe("Additional workspace directories passed as --add-dir"),
             correlationId: z.string().optional().describe("Request trace ID (auto if omitted)"),
             optimizePrompt: z.boolean().default(false).describe("Optimize prompt before execution"),
             idleTimeoutMs: z
@@ -4890,19 +4725,19 @@ export function createGatewayServer(deps = {}) {
             outputFormat: z
                 .enum(["text", "json", "stream-json"])
                 .default("text")
-                .describe("Gemini output format. `json` emits `-o json` (single JSON with usageMetadata). `stream-json` emits `-o stream-json` (NDJSON event stream — `init`/`message`/`result` lines, usage extracted from the terminal `result.stats` event). Both report usage to the flight recorder."),
-            sandbox: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.sandbox.describe("Run Gemini in sandbox mode (-s)"),
-            policyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.policyFiles.describe("Policy file paths (--policy <path>, one per file). Paths must exist."),
-            adminPolicyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.adminPolicyFiles.describe("Admin policy file paths (--admin-policy <path>, one per file). Paths must exist."),
-            attachments: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.attachments.describe("Absolute file paths prepended as @<path> tokens to the prompt"),
+                .describe("Antigravity CLI currently supports text output only through the gateway; json and stream-json are rejected."),
+            sandbox: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.sandbox.describe("Run Antigravity in sandbox mode (--sandbox)"),
+            policyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.policyFiles.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
+            adminPolicyFiles: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.adminPolicyFiles.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
+            attachments: GEMINI_HIGH_IMPACT_PARAMS_SCHEMA.shape.attachments.describe("Unsupported for Antigravity CLI; non-empty values are rejected."),
             skipTrust: z
                 .boolean()
                 .default(false)
-                .describe("Emit `--skip-trust` so Gemini trusts the workspace for this session and skips the interactive trust prompt (Phase 4 slice γ). Required for headless runs in fresh workspaces."),
+                .describe("Unsupported for Antigravity CLI; true is rejected."),
             yolo: z
                 .boolean()
                 .optional()
-                .describe("Emit `--yolo` to auto-approve all actions. Equivalent to approvalMode 'yolo'; routed through the same approval gate. Under mcp_managed the gate still decides."),
+                .describe("Emit `--dangerously-skip-permissions` to auto-approve all actions. Routed through the same approval gate. Under mcp_managed the gate still decides."),
             workspace: WORKSPACE_ALIAS_SCHEMA.optional(),
             worktree: WORKTREE_SCHEMA.optional(),
         }, {