@caveman-ai/cli 2.0.0 → 2.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -37,6 +37,14 @@ function compatTarget(base, path) {
37
37
  while (overlap > 0 && left.slice(-overlap).join("/") !== right.slice(0, overlap).join("/")) overlap--;
38
38
  return base.origin + "/" + [...left, ...right.slice(overlap)].join("/");
39
39
  }
40
+ function verifiedChatGPTSubscriptionRoute(gateway, api, provider, originalBaseUrl, published) {
41
+ if (api !== "openai-codex-responses" || provider !== "openai-codex") return void 0;
42
+ const original = endpoint(originalBaseUrl);
43
+ const upstream = endpoint(published?.provider_upstreams?.["openai-codex"]);
44
+ if (!original || !upstream || !endpoint(gateway)) return void 0;
45
+ if (append(upstream, "/codex/responses") !== append(original, "/codex/responses")) return void 0;
46
+ return trimTrailingSlashes(gateway);
47
+ }
40
48
  function verifiedProviderRoute(gateway, api, provider, originalBaseUrl, published) {
41
49
  const original = endpoint(originalBaseUrl);
42
50
  if (!original || !endpoint(gateway)) return void 0;
@@ -122,11 +130,36 @@ import { dirname, extname, isAbsolute, join, resolve } from "node:path";
122
130
  function parseWindowsNodeShim(source) {
123
131
  for (const line of source.split(/\r?\n/)) {
124
132
  if (!/(?:\bnode(?:\.exe)?\b|_prog)/i.test(line) || !/%\*/.test(line)) continue;
125
- const match = line.match(/"%(?:dp0%|~dp0)\\([^"\r\n]+\.(?:cjs|mjs|js))"\s+%\*/i) ?? line.match(/"([A-Za-z]:[\\/][^"\r\n]+\.(?:cjs|mjs|js))"\s+%\*/i);
133
+ const match = line.match(/"%(?:dp0%|~dp0)\\?([^"\r\n]+\.(?:cjs|mjs|js))"\s+%\*/i) ?? line.match(/"([A-Za-z]:[\\/][^"\r\n]+\.(?:cjs|mjs|js))"\s+%\*/i);
126
134
  if (match) return match[1];
127
135
  }
128
136
  return null;
129
137
  }
138
+ function parseWindowsNestedShim(source) {
139
+ const match = source.match(/^(?:@echo[ \t]+off[ \t]*\r?\n)?[ \t]*"%~dp0\\?([\w .@+\-\\/]+\.(?:cmd|bat))"[ \t]+%\*[ \t]*(?:\r?\n)?$/i);
140
+ return match?.[1] ?? null;
141
+ }
142
+ function resolveWindowsNodeShim(executable, depth = 0, seen = /* @__PURE__ */ new Set()) {
143
+ if (depth > 4) throw new Error("Windows command shim nesting too deep");
144
+ const normalized = resolve(executable);
145
+ if (seen.has(normalized.toLowerCase())) throw new Error("Windows command shim cycle");
146
+ seen.add(normalized.toLowerCase());
147
+ const stat = statSync(normalized);
148
+ if (!stat.isFile() || stat.size > 256 * 1024) {
149
+ throw new Error(`cannot safely launch Windows command shim: ${normalized}`);
150
+ }
151
+ const source = readFileSync(normalized, "utf8");
152
+ const nested = parseWindowsNestedShim(source);
153
+ const forwardsToBatch = /"[^"\r\n]+\.(?:cmd|bat)"[ \t]+%\*/i.test(source);
154
+ const jsTarget = forwardsToBatch ? null : parseWindowsNodeShim(source);
155
+ const child = jsTarget ?? nested;
156
+ if (!child) {
157
+ throw new Error(`cannot safely launch non-Node Windows command shim: ${normalized}; install a native .exe`);
158
+ }
159
+ const target = /^[A-Za-z]:[\\/]/.test(child) ? child : resolve(dirname(normalized), ...child.split(/[\\/]+/));
160
+ if (!existsSync(target) || !statSync(target).isFile()) throw new Error(`Windows command shim target is missing: ${target}`);
161
+ return jsTarget ? target : resolveWindowsNodeShim(target, depth + 1, seen);
162
+ }
130
163
  function envValue(env, name) {
131
164
  const key = Object.keys(env).find((candidate) => candidate.toLowerCase() === name.toLowerCase());
132
165
  return key === void 0 ? void 0 : env[key];
@@ -151,18 +184,7 @@ function portableInvocation(command, args, platform = process.platform, env = pr
151
184
  if (platform !== "win32") return { command, args: [...args] };
152
185
  const executable = resolveWindowsCommand(command, env) ?? command;
153
186
  if (!/\.(?:cmd|bat)$/i.test(executable)) return { command: executable, args: [...args] };
154
- const stat = statSync(executable);
155
- if (!stat.isFile() || stat.size > 256 * 1024) {
156
- throw new Error(`cannot safely launch Windows command shim: ${executable}`);
157
- }
158
- const shimScript = parseWindowsNodeShim(readFileSync(executable, "utf8"));
159
- if (!shimScript) {
160
- throw new Error(`cannot safely launch non-Node Windows command shim: ${executable}; install a native .exe`);
161
- }
162
- const script = /^[A-Za-z]:[\\/]/.test(shimScript) ? shimScript : resolve(dirname(executable), ...shimScript.split(/[\\/]+/));
163
- if (!statSync(script).isFile()) {
164
- throw new Error(`Windows command shim target is missing: ${script}`);
165
- }
187
+ const script = resolveWindowsNodeShim(executable);
166
188
  return { command: process.execPath, args: [script, ...args] };
167
189
  }
168
190
 
@@ -205,6 +227,14 @@ function outputReplacementOf(response) {
205
227
  const replacement = response?.hookSpecificOutput?.updatedToolOutput ?? response?.output_replacement;
206
228
  return replacement ? replacement : void 0;
207
229
  }
230
+ function normalizeRecoveryHandle(value) {
231
+ if (!fits(value, MAX_RECOVERY_REF_BYTES)) return void 0;
232
+ let handle = value.trim();
233
+ if (handle.startsWith("<<") && handle.endsWith(">>")) handle = handle.slice(2, -2).trim();
234
+ if (handle.startsWith("ccr://")) handle = handle.slice(6).trim().replace(/^\/+|\/+$/g, "");
235
+ else if (handle.startsWith("ccr:")) handle = handle.slice(4).trim();
236
+ return /^ccr_[A-Za-z0-9_-]+$/.test(handle) ? handle : void 0;
237
+ }
208
238
  var ROUTES_BY_API = {
209
239
  "anthropic-messages": "/w/pi",
210
240
  "openai-completions": "/w/pi/openai/v1",
@@ -212,6 +242,9 @@ var ROUTES_BY_API = {
212
242
  "google-generative-ai": "/w/pi/v1beta"
213
243
  };
214
244
  var ROUTES_BY_PROVIDER_API = {
245
+ "openai-codex": {
246
+ "openai-codex-responses": "/w/pi"
247
+ },
215
248
  "opencode-go": {
216
249
  "anthropic-messages": "/w/pi/compat/opencode-go",
217
250
  "openai-completions": "/w/pi/compat/opencode-go/v1",
@@ -222,6 +255,7 @@ var UPSTREAM_HOSTS_BY_PROVIDER = {
222
255
  anthropic: "api.anthropic.com",
223
256
  openai: "api.openai.com",
224
257
  google: "generativelanguage.googleapis.com",
258
+ "openai-codex": "chatgpt.com",
225
259
  "opencode-go": "opencode.ai"
226
260
  };
227
261
  function upstreamHostFor(provider) {
@@ -243,13 +277,8 @@ function compatUpstreamFor(provider, compatUpstreams) {
243
277
  }
244
278
  function routeForApi(gateway, api, provider, originalBaseUrl, compatUpstreams, providerUpstreams) {
245
279
  if (originalBaseUrl !== void 0) {
246
- return verifiedProviderRoute(
247
- joinUrl(gateway, "/w/pi"),
248
- api,
249
- provider ?? "",
250
- originalBaseUrl,
251
- { compat_upstreams: compatUpstreams, provider_upstreams: providerUpstreams }
252
- );
280
+ const published = { compat_upstreams: compatUpstreams, provider_upstreams: providerUpstreams };
281
+ return verifiedChatGPTSubscriptionRoute(joinUrl(gateway, "/w/pi"), api, provider ?? "", originalBaseUrl, published) ?? verifiedProviderRoute(joinUrl(gateway, "/w/pi"), api, provider ?? "", originalBaseUrl, published);
253
282
  }
254
283
  const table = (provider ? ROUTES_BY_PROVIDER_API[provider] : void 0) ?? ROUTES_BY_API;
255
284
  if (provider && !UPSTREAM_HOSTS_BY_PROVIDER[provider]) return void 0;
@@ -413,6 +442,15 @@ function compatForRoutedModel(model) {
413
442
  const explicit2 = model.compat;
414
443
  return { ...explicit2, sessionAffinityFormat: explicit2?.sessionAffinityFormat ?? (openRouter ? "openrouter" : "openai") };
415
444
  }
445
+ if (model.api === "anthropic-messages") {
446
+ if (!openRouter) return model.compat;
447
+ const explicit2 = model.compat;
448
+ return {
449
+ ...explicit2,
450
+ sendSessionAffinityHeaders: explicit2?.sendSessionAffinityHeaders ?? true,
451
+ sessionAffinityFormat: explicit2?.sessionAffinityFormat ?? "openrouter"
452
+ };
453
+ }
416
454
  if (model.api !== "openai-completions") return model.compat;
417
455
  const explicit = model.compat;
418
456
  const zai = provider === "zai" || provider === "zai-coding-cn" || url.includes("api.z.ai") || url.includes("open.bigmodel.cn");
@@ -435,7 +473,7 @@ function compatForRoutedModel(model) {
435
473
  maxTokensField: explicit?.maxTokensField ?? (maxTokens ? "max_tokens" : "max_completion_tokens"),
436
474
  requiresReasoningContentOnAssistantMessages: explicit?.requiresReasoningContentOnAssistantMessages ?? deepseek,
437
475
  thinkingFormat: explicit?.thinkingFormat ?? (deepseek ? "deepseek" : zai ? "zai" : together ? "together" : antLing ? "ant-ling" : openRouter ? "openrouter" : "openai"),
438
- supportsStrictMode: explicit?.supportsStrictMode ?? !(moonshot || together || cloudflareGateway || nvidia),
476
+ sendSessionAffinityHeaders: explicit?.sendSessionAffinityHeaders ?? openRouter,
439
477
  sessionAffinityFormat: explicit?.sessionAffinityFormat ?? (openRouter ? "openrouter" : "openai"),
440
478
  supportsLongCacheRetention: explicit?.supportsLongCacheRetention ?? !(together || cloudflareWorkers || cloudflareGateway || nvidia || antLing)
441
479
  };
@@ -495,11 +533,15 @@ var ProviderRouter = class {
495
533
  const compatibilityIssue = model.api === "openai-completions" && original.includes("api.openai.com") ? "Pi cannot preserve this OpenAI Chat endpoint's prompt cache keys through routing" : void 0;
496
534
  let headerIssue;
497
535
  let oauth = true;
536
+ let authKnown = false;
498
537
  try {
499
538
  oauth = ctx.modelRegistry.isUsingOAuth(model);
539
+ authKnown = true;
500
540
  } catch {
501
541
  }
502
- if (!oauth && route && !compatibilityIssue) {
542
+ const chatGPTSubscriptionModel = model.provider === "openai-codex" && model.api === "openai-codex-responses";
543
+ const authIssue = !authKnown ? "authentication type could not be verified" : oauth && !chatGPTSubscriptionModel ? "OAuth/subscription credentials are not routed" : !oauth && chatGPTSubscriptionModel && route ? "ChatGPT subscription route requires OAuth" : void 0;
544
+ if (!oauth && route && !compatibilityIssue && !chatGPTSubscriptionModel) {
503
545
  try {
504
546
  const auth = await ctx.modelRegistry.getApiKeyAndHeaders(model);
505
547
  if (!auth.ok) throw new Error("request auth unavailable");
@@ -530,13 +572,13 @@ var ProviderRouter = class {
530
572
  }
531
573
  if (!this.gateOpen || this.gateGeneration !== gateGeneration || ctx.model !== model) return;
532
574
  }
533
- if (!route || oauth || compatibilityIssue || headerIssue) {
575
+ if (!route || authIssue || compatibilityIssue || headerIssue) {
534
576
  if (!await this.restoreCurrentModel(ctx)) return;
535
577
  if (!this.warnedModels.has(key)) {
536
578
  this.warnedModels.add(key);
537
579
  const mount = compatUpstreamFor(model.provider, this.compatUpstreams);
538
580
  const expected = mount !== void 0 ? hostOf(mount) : upstreamHostFor(model.provider);
539
- const reason = oauth ? "OAuth/subscription credentials are not routed" : compatibilityIssue ?? headerIssue ?? (expected === void 0 ? `no compat mount named "${model.provider}" in the local proxy; add compat.${model.provider}.base_url to caveman.yaml to route it` : hostOf(original) !== expected ? `provider endpoint ${hostOf(original) ?? original} is not ${expected}` : `provider endpoint path or API "${model.provider}/${model.api}" is not verified by the running proxy`);
581
+ const reason = authIssue ?? compatibilityIssue ?? headerIssue ?? (expected === void 0 ? `no compat mount named "${model.provider}" in the local proxy; add compat.${model.provider}.base_url to caveman.yaml to route it` : hostOf(original) !== expected ? `provider endpoint ${hostOf(original) ?? original} is not ${expected}` : `provider endpoint path or API "${model.provider}/${model.api}" is not verified by the running proxy`);
540
582
  this.notify(boundedString(`Caveman: pass-through for ${key} (${reason}); no compression`, MAX_MESSAGE_BYTES), "warning");
541
583
  }
542
584
  return;
@@ -598,12 +640,14 @@ var ProviderRouter = class {
598
640
 
599
641
  // ../pi-extension/src/recovery.ts
600
642
  import { execFile as execFile2, spawn } from "node:child_process";
643
+ import { createHash as createHash2 } from "node:crypto";
601
644
  import { existsSync as existsSync2 } from "node:fs";
602
645
  import { homedir as homedir2 } from "node:os";
603
646
  import { delimiter, join as join3 } from "node:path";
604
647
  var PROBE_TIMEOUT_MS = 2e3;
605
648
  var CALL_TIMEOUT_MS = 3e4;
606
649
  var INIT_TIMEOUT_MS = 2e3;
650
+ var VERIFY_TIMEOUT_MS = 2e3;
607
651
  function resolveMcpBinary(env = process.env) {
608
652
  const explicit = env.CAVEMAN_MCP_BIN?.trim();
609
653
  if (explicit) return explicit;
@@ -619,6 +663,7 @@ var RecoveryClient = class {
619
663
  child;
620
664
  initialized = false;
621
665
  probed;
666
+ verificationSupported = false;
622
667
  ensuring;
623
668
  nextId = 1;
624
669
  pending = /* @__PURE__ */ new Map();
@@ -662,6 +707,28 @@ var RecoveryClient = class {
662
707
  return { text: JSON.stringify({ error: "cave_recovery_transport", message: error.message }), isError: true };
663
708
  }
664
709
  }
710
+ // A silent pre-publication proof must not count as delivery to the model.
711
+ // Old companions ignore unknown arguments, so capability proof is mandatory
712
+ // before sending verify_only through the same long-lived recovery child.
713
+ async verify(handle, originalText) {
714
+ const normalized = normalizeRecoveryHandle(handle);
715
+ const bytes = Buffer.byteLength(originalText, "utf8");
716
+ if (!normalized || bytes > MAX_TOOL_OUTPUT_BYTES) return false;
717
+ try {
718
+ if (!await this.ensure() || !this.verificationSupported) return false;
719
+ const result = await this.call("tools/call", {
720
+ name: "caveman_retrieve",
721
+ arguments: { recovery_handle: normalized, verify_only: true }
722
+ }, void 0, VERIFY_TIMEOUT_MS);
723
+ if (result.isError || !Array.isArray(result.content) || result.content.length !== 1) return false;
724
+ const block = result.content[0];
725
+ if (block?.type !== "text" || typeof block.text !== "string" || Buffer.byteLength(block.text, "utf8") > MAX_MESSAGE_BYTES) return false;
726
+ const proof = JSON.parse(block.text);
727
+ return proof?.recovery_handle === normalized && proof.byte_length === bytes && proof.sha256 === createHash2("sha256").update(originalText, "utf8").digest("hex");
728
+ } catch {
729
+ return false;
730
+ }
731
+ }
665
732
  dispose() {
666
733
  this.disposed = true;
667
734
  const child = this.child;
@@ -699,6 +766,7 @@ var RecoveryClient = class {
699
766
  if (error) return resolve2(false);
700
767
  try {
701
768
  const parsed = JSON.parse(stdout);
769
+ this.verificationSupported = Array.isArray(parsed?.capabilities) && parsed.capabilities.includes("recovery_verification");
702
770
  resolve2(Array.isArray(parsed?.capabilities) && parsed.capabilities.includes("mcp_recovery"));
703
771
  } catch {
704
772
  resolve2(false);
@@ -727,7 +795,7 @@ var RecoveryClient = class {
727
795
  await this.call("initialize", {
728
796
  protocolVersion: "2024-11-05",
729
797
  capabilities: {},
730
- clientInfo: { name: "caveman-pi-extension", version: "0.2.0" }
798
+ clientInfo: { name: "caveman-pi-extension", version: "0.3.0" }
731
799
  }, void 0, INIT_TIMEOUT_MS);
732
800
  this.notify("notifications/initialized");
733
801
  if (this.disposed) {
@@ -845,18 +913,35 @@ var RecoveryClient = class {
845
913
  };
846
914
 
847
915
  // ../pi-extension/src/tool-output.ts
848
- async function shrinkToolResult(bridge, sessionId, event) {
916
+ async function shrinkToolResult(bridge, sessionId, event, recovery) {
849
917
  const text = event.content.map((block) => block.type === "text" && typeof block.text === "string" ? block.text : "").join("");
850
918
  if (!text) return void 0;
851
919
  if (Buffer.byteLength(text, "utf8") > MAX_TOOL_OUTPUT_BYTES) return void 0;
852
- const response = await bridge.call(event.isError ? "PostToolUseFailure" : "PostToolUse", {
853
- session_id: sessionId,
854
- tool_name: event.toolName,
855
- tool_input: event.input,
856
- tool_output: text
857
- });
858
- const replacement = outputReplacementOf(response);
859
- if (!replacement) return void 0;
920
+ let replacement;
921
+ try {
922
+ const response = await bridge.call(event.isError ? "PostToolUseFailure" : "PostToolUse", {
923
+ session_id: sessionId,
924
+ tool_name: event.toolName,
925
+ tool_input: event.input,
926
+ tool_output: text
927
+ });
928
+ const proposed = outputReplacementOf(response);
929
+ if (!proposed || Buffer.byteLength(proposed, "utf8") > MAX_OUTPUT_REPLACEMENT_BYTES) return void 0;
930
+ let handle = normalizeRecoveryHandle(response?.recovery_ref);
931
+ if (response?.recovery_ref !== void 0 && !handle) return void 0;
932
+ let advertised = false;
933
+ for (const match of proposed.matchAll(/(?:<<)?ccr(?::|_)[^\s"'`]*/g)) {
934
+ if (match.index > 0 && /[A-Za-z0-9_<:/]/.test(proposed[match.index - 1])) return void 0;
935
+ const candidate = normalizeRecoveryHandle(match[0]);
936
+ if (!candidate || handle && candidate !== handle) return void 0;
937
+ handle = candidate;
938
+ advertised = true;
939
+ }
940
+ if (!advertised || !handle || !await recovery.verify(handle, text)) return void 0;
941
+ replacement = proposed;
942
+ } catch {
943
+ return void 0;
944
+ }
860
945
  return { content: [{ type: "text", text: replacement }, ...event.content.filter((block) => block.type !== "text")] };
861
946
  }
862
947
 
@@ -1053,7 +1138,7 @@ function index_default(pi) {
1053
1138
  pendingBytes += bytes;
1054
1139
  }
1055
1140
  }));
1056
- pi.on("tool_result", GUARD_tool_result((event) => event.toolName === RECOVERY_TOOL ? void 0 : shrinkToolResult(bridge, sessionId, event)));
1141
+ pi.on("tool_result", GUARD_tool_result((event) => event.toolName === RECOVERY_TOOL ? void 0 : shrinkToolResult(bridge, sessionId, event, recovery)));
1057
1142
  pi.on("session_before_compact", GUARD_session_before_compact(() => {
1058
1143
  void bridge.call("PreCompact", { session_id: sessionId });
1059
1144
  }));