opencrowd 0.15.0 → 0.15.2

package/bundle/opencrowd.js

@@ -286,13 +286,13 @@ async function listArtifacts(state, prefix = "") {
   const { readdir: readdir3 } = await import("node:fs/promises");
   const start = await safeArtifactPath(state, prefix || ".");
   const files = [];
-  await walk3(start, files, state.artifactsDir);
+  await walk4(start, files, state.artifactsDir);
   return files.sort();
-  async function walk3(dir, out, root) {
+  async function walk4(dir, out, root) {
     for (const entry of await readdir3(dir, { withFileTypes: true })) {
       const full = join3(dir, entry.name);
       if (entry.isDirectory()) {
-        await walk3(full, out, root);
+        await walk4(full, out, root);
       } else if (entry.isFile()) {
         out.push(relative2(root, full));
       }
@@ -15903,18 +15903,45 @@ var init_redact = __esm({
   }
 });
 
+// ../../packages/protocol/dist/limits.js
+function capStrings(value, maxChars = MAX_EXTERNAL_STRING_CHARS, depth = 0) {
+  if (typeof value === "string")
+    return capString(value, maxChars);
+  if (depth > 30 || !value || typeof value !== "object")
+    return value;
+  if (Array.isArray(value))
+    return value.map((item) => capStrings(item, maxChars, depth + 1));
+  return Object.fromEntries(Object.entries(value).map(([key, item]) => [key, capStrings(item, maxChars, depth + 1)]));
+}
+function capString(value, maxChars = MAX_EXTERNAL_STRING_CHARS) {
+  return value.length <= maxChars ? value : `${value.slice(0, maxChars)}[truncated ${value.length - maxChars} chars]`;
+}
+var MAX_EXTERNAL_STRING_CHARS;
+var init_limits = __esm({
+  "../../packages/protocol/dist/limits.js"() {
+    "use strict";
+    MAX_EXTERNAL_STRING_CHARS = 128 * 1024;
+  }
+});
+
 // ../../packages/protocol/dist/telemetry.js
 function scrubText(text) {
   return text.replace(/-----BEGIN [^-]*PRIVATE KEY-----[\s\S]*?-----END [^-]*PRIVATE KEY-----/g, "[REDACTED]").replace(/\bBearer\s+[A-Za-z0-9._~+\/-]+=*/gi, "Bearer [REDACTED]").replace(/\b(?:sk-[A-Za-z0-9_-]{16,}|ph[csx]_[A-Za-z0-9_-]{16,}|gh[pousr]_[A-Za-z0-9]{20,}|eyJ[A-Za-z0-9_-]+\.[A-Za-z0-9_-]+\.[A-Za-z0-9_-]+)\b/g, "[REDACTED]").replace(/\b((?:[A-Z_]*(?:SECRET|PASSWORD|API_KEY|PRIVATE_KEY|ACCESS_TOKEN)[A-Z_]*|authorization|cookie|signature|x-amz-signature)\s*[=:]\s*)[^\s&,;]+/gi, "$1[REDACTED]").replace(/(https?:\/\/)[^\s/@]+:[^\s/@]+@/g, "$1[REDACTED]@");
 }
 function captureDetail(input2, maxBytes = 96 * 1024) {
   let redacted = false;
+  let spent = 0, truncatedStrings = false;
+  const budget = maxBytes * 8;
   const clean3 = (value2, depth = 0) => {
     if (depth > 30)
       return "[depth limit]";
     if (typeof value2 === "string") {
-      const s3 = scrubText(value2);
-      redacted ||= s3 !== value2;
+      const limit = Math.min(maxBytes, Math.max(0, budget - spent));
+      truncatedStrings ||= value2.length > limit;
+      const capped = capString(value2, limit);
+      spent += capped.length;
+      const s3 = scrubText(capped);
+      redacted ||= s3 !== capped;
       return s3;
     }
     if (Array.isArray(value2))
@@ -15934,13 +15961,14 @@ function captureDetail(input2, maxBytes = 96 * 1024) {
   return {
     value: bytes > maxBytes ? { preview: json3.slice(0, Math.floor(maxBytes / 6)) } : value,
     redacted,
-    truncated: bytes > maxBytes,
+    truncated: bytes > maxBytes || truncatedStrings,
     bytes
   };
 }
 var init_telemetry = __esm({
   "../../packages/protocol/dist/telemetry.js"() {
     "use strict";
+    init_limits();
     init_redact();
   }
 });
@@ -16209,6 +16237,7 @@ var init_dist = __esm({
     init_commands();
     init_events();
     init_redact();
+    init_limits();
     init_telemetry();
     init_seq();
     init_codec();
@@ -58334,30 +58363,110 @@ var init_crowdcode_preferences = __esm({
 // ../../packages/economy/dist/media.js
 import { createHash as createHash2 } from "node:crypto";
 async function extractMedia(session, value, depth = 0) {
-  if (depth > 20 || !value || typeof value !== "object")
+  return walk3(session, value, void 0, void 0, depth);
+}
+async function walk3(session, value, key, siblings, depth) {
+  if (depth > 20 || !value)
+    return value;
+  if (typeof value === "string")
+    return await extractString(session, value, key, siblings) ?? value;
+  if (typeof value !== "object")
     return value;
   if (Array.isArray(value))
-    return Promise.all(value.map((item) => extractMedia(session, item, depth + 1)));
+    return Promise.all(value.map((item) => walk3(session, item, key, siblings, depth + 1)));
   const source = value;
-  const extensions = { "video/mp4": "mp4", "video/webm": "webm", "image/png": "png", "image/jpeg": "jpg", "image/webp": "webp", "audio/mpeg": "mp3", "audio/wav": "wav" };
-  const mime = String(source.media_type ?? source.mime_type ?? "image/png");
-  if (typeof source.b64_json === "string" && extensions[mime] && /^[A-Za-z0-9+/\r\n]*={0,2}$/.test(source.b64_json)) {
-    const bytes = Buffer.from(source.b64_json, "base64");
-    if (bytes.length > 24 * 1024 * 1024) {
-      const { b64_json: _inline2, ...metadata2 } = source;
-      return { ...metadata2, media_error: "Generated media exceeds 24 MiB artifact limit", bytes: bytes.length };
-    }
-    const name = createHash2("sha256").update(bytes).digest("hex").slice(0, 20);
-    const artifact = await saveArtifact(session, `generated/${name}.${extensions[mime]}`, bytes, { media_type: mime });
-    const { b64_json: _inline, ...metadata } = source;
-    return { ...metadata, artifact_path: artifact.path, bytes: bytes.length };
+  const entries = await Promise.all(Object.entries(source).map(async ([name, item]) => [name, await walk3(session, item, name, source, depth + 1)]));
+  const openAiStyle = entries.find(([name, item]) => name === "b64_json" && isReplacement(item));
+  if (!openAiStyle)
+    return Object.fromEntries(entries);
+  return { ...Object.fromEntries(entries.filter(([name]) => name !== "b64_json")), ...openAiStyle[1] };
+}
+function isReplacement(value) {
+  return Boolean(value && typeof value === "object" && ("artifact_path" in value || "media_error" in value));
+}
+async function extractString(session, value, key, siblings) {
+  const payload = dataUriPayload(value) ?? { base64: value, declared: declaredMediaType(key, siblings) };
+  if (payload.base64.length < MIN_BASE64_CHARS || !looksLikeBase64(payload.base64))
+    return void 0;
+  const mediaType = sniff(Buffer.from(payload.base64.slice(0, SNIFF_CHARS), "base64")) ?? payload.declared;
+  if (!mediaType || !EXTENSIONS[mediaType] || !isCompleteBase64(payload.base64))
+    return void 0;
+  const bytes = Buffer.from(payload.base64, "base64");
+  if (bytes.length > MAX_ARTIFACT_BYTES) {
+    return { media_error: `Generated media exceeds ${MAX_ARTIFACT_BYTES / (1024 * 1024)} MiB artifact limit`, bytes: bytes.length, media_type: mediaType };
   }
-  return Object.fromEntries(await Promise.all(Object.entries(source).map(async ([key, item]) => [key, await extractMedia(session, item, depth + 1)])));
+  const name = createHash2("sha256").update(bytes).digest("hex").slice(0, 20);
+  const artifact = await saveArtifact(session, `generated/${name}.${EXTENSIONS[mediaType]}`, bytes, { media_type: mediaType });
+  return { artifact_path: artifact.path, bytes: bytes.length, media_type: mediaType };
+}
+function declaredMediaType(key, siblings) {
+  if (key !== "b64_json")
+    return void 0;
+  const declared = siblings?.media_type ?? siblings?.mime_type;
+  return typeof declared === "string" ? declared : "image/png";
+}
+function dataUriPayload(value) {
+  const match = /^data:([a-z]+\/[a-z0-9.+-]+)?(?:;[a-z-]+=[^;,]*)*;base64,/i.exec(value);
+  return match ? { base64: value.slice(match[0].length), declared: match[1]?.toLowerCase() } : void 0;
+}
+function looksLikeBase64(value) {
+  return /^[A-Za-z0-9+/_-]{64}/.test(value);
+}
+function isCompleteBase64(value) {
+  return /^[A-Za-z0-9+/_\-\s]*={0,2}$/.test(value);
+}
+function sniff(bytes) {
+  const ascii = (at, text) => bytes.length >= at + text.length && bytes.subarray(at, at + text.length).toString("latin1") === text;
+  if (ascii(0, "RIFF"))
+    return ascii(8, "WAVE") ? "audio/wav" : ascii(8, "WEBP") ? "image/webp" : ascii(8, "AVI ") ? "video/x-msvideo" : void 0;
+  if (ascii(0, "ID3"))
+    return "audio/mpeg";
+  if (ascii(0, "OggS"))
+    return "audio/ogg";
+  if (ascii(0, "fLaC"))
+    return "audio/flac";
+  if (ascii(0, "%PDF-"))
+    return "application/pdf";
+  if (ascii(0, "GIF87a") || ascii(0, "GIF89a"))
+    return "image/gif";
+  if (bytes.length >= 8 && bytes.subarray(0, 8).equals(Buffer.from([137, 80, 78, 71, 13, 10, 26, 10])))
+    return "image/png";
+  if (bytes.length >= 3 && bytes[0] === 255 && bytes[1] === 216 && bytes[2] === 255)
+    return "image/jpeg";
+  if (bytes.length >= 4 && bytes.subarray(0, 4).equals(Buffer.from([26, 69, 223, 163])))
+    return "video/webm";
+  if (ascii(4, "ftyp")) {
+    const brand = bytes.subarray(8, 12).toString("latin1");
+    return brand.startsWith("qt") ? "video/quicktime" : /^(M4A|M4B|mp4a)/.test(brand) ? "audio/mp4" : "video/mp4";
+  }
+  if (bytes.length >= 4 && bytes[0] === 255 && (bytes[1] & 224) === 224 && (bytes[1] & 24) !== 8 && (bytes[1] & 6) !== 0 && (bytes[2] & 240) !== 240 && (bytes[2] & 240) !== 0)
+    return "audio/mpeg";
+  return void 0;
 }
+var MAX_ARTIFACT_BYTES, MIN_BASE64_CHARS, SNIFF_CHARS, EXTENSIONS;
 var init_media = __esm({
   "../../packages/economy/dist/media.js"() {
     "use strict";
     init_dist2();
+    MAX_ARTIFACT_BYTES = 24 * 1024 * 1024;
+    MIN_BASE64_CHARS = 1024;
+    SNIFF_CHARS = 64;
+    EXTENSIONS = {
+      "video/mp4": "mp4",
+      "video/webm": "webm",
+      "video/quicktime": "mov",
+      "video/x-msvideo": "avi",
+      "image/png": "png",
+      "image/jpeg": "jpg",
+      "image/gif": "gif",
+      "image/webp": "webp",
+      "audio/mpeg": "mp3",
+      "audio/wav": "wav",
+      "audio/ogg": "ogg",
+      "audio/flac": "flac",
+      "audio/mp4": "m4a",
+      "application/pdf": "pdf"
+    };
   }
 });
 
@@ -58474,6 +58583,9 @@ var init_purchases = __esm({
 // ../../packages/economy/dist/gateway.js
 import { readFile as readFile10, writeFile as writeFile5 } from "node:fs/promises";
 import { join as join12 } from "node:path";
+async function boundResponse(session, data) {
+  return capStrings(await extractMedia(session, data), MAX_EXTERNAL_STRING_CHARS);
+}
 function requestFor(endpoint, method, body) {
   if (method !== "GET" && method !== "HEAD")
     return { url: endpoint, body };
@@ -58568,6 +58680,7 @@ var init_gateway = __esm({
     "use strict";
     init_crowdcode_preferences();
     init_media();
+    init_dist();
     init_dist2();
     init_offers();
     init_approvals();
@@ -58651,7 +58764,7 @@ var init_gateway = __esm({
             case "review_paid_service":
               return await this.reviewPaidService(args);
             case "read_service":
-              return this.options.agentcash.read ? await this.options.agentcash.read(String(args.url ?? "")) : { ok: false, error: "Authenticated reads are unavailable" };
+              return await this.readService(args);
             case "bridge_usdc":
               return await this.bridgeUsdc(args, operationId);
             default:
@@ -58695,6 +58808,15 @@ var init_gateway = __esm({
       async serviceEvidence(endpoint) {
         return await this.crowdcodeEnabled() ? this.options.crowdcode.getServiceScore({ apiEndpoint: endpoint }) : { ok: true, unproven: true, summary: "CrowdCode is off; reputation checks and submissions are disabled." };
       }
+      /** An unpaid read is still an external body: bound it like a paid one. */
+      async readService(args) {
+        if (!this.options.agentcash.read)
+          return { ok: false, error: "Authenticated reads are unavailable" };
+        const result = await this.options.agentcash.read(String(args.url ?? ""));
+        if (!result.ok || result.data === void 0)
+          return result;
+        return { ...result, data: await boundResponse(this.options.session, result.data) };
+      }
       async getWalletStatus() {
         const result = await this.options.agentcash.getBalance();
         if (!result.ok) {
@@ -58851,7 +58973,7 @@ var init_gateway = __esm({
         const paid = outcome === "paid_success" || outcome === "paid_failure";
         const chargedCents = result.ambiguous ? quotedCostCents : paid ? Math.max(1, Math.round((result.payment?.paidUsd ?? quotedCostCents / 100) * 100)) : 0;
         if (result.data)
-          result.data = await extractMedia(session, result.data);
+          result.data = await boundResponse(session, result.data);
         const artifact = result.ambiguityReason === "transport" ? void 0 : await saveArtifact(session, `service-calls/${Date.now()}-${slugUrl(endpoint)}.json`, JSON.stringify({ status: result.status, body: result.data }, null, 2), { endpoint, method });
         const purchaseId = newPurchaseId();
         const railUsed = result.payment?.rail ?? inspection.rail;
@@ -59369,6 +59491,9 @@ var init_mocks = __esm({
           }
         };
       }
+      async read(url3) {
+        return url3 in (this.options.readResults ?? {}) ? { ok: true, data: this.options.readResults[url3] } : { ok: false, error: `no scripted read for ${url3}` };
+      }
       async bridge(input2) {
         return { ok: true, data: { bridged: input2.amountUsd, from: input2.from, to: input2.to, mock: true } };
       }
@@ -59483,6 +59608,7 @@ var init_dist4 = __esm({
     init_crowdcode();
     init_gateway();
     init_mcp();
+    init_media();
     init_monad();
     init_mocks();
     init_purchases();
@@ -87058,7 +87184,9 @@ function createHostedToolExecutor(options) {
     }
     try {
       const result = await postHostedTool(options, name, { ...args, ...source === void 0 ? {} : { source } }, { signal: context.signal });
-      return result.ok ? { ok: true, data: result.data } : { ok: false, error: result.error ?? "tool failed" };
+      if (result.ok)
+        return { ok: true, data: await boundResponse(context.session, result.data) };
+      return { ok: false, error: result.error ?? "tool failed" };
     } catch (error53) {
       if (context.signal?.aborted)
         throw error53;
@@ -87074,6 +87202,7 @@ var init_hosted_provider = __esm({
     "use strict";
     init_dist();
     init_dist2();
+    init_dist4();
     init_worker();
     HostedRequestError = class extends Error {
       code;
@@ -87703,7 +87832,7 @@ This artifact was created by the real worker using a scripted model. No USDC was
     }
   };
 }
-var RuntimePause, MachineWorker;
+var MAX_WORKER_STATE_CHARS, RuntimePause, MachineWorker;
 var init_worker = __esm({
   "../../packages/agent-runtime/dist/worker.js"() {
     "use strict";
@@ -87714,6 +87843,7 @@ var init_worker = __esm({
     init_dist6();
     init_hosted_economy();
     init_knowledge();
+    MAX_WORKER_STATE_CHARS = 2 * 1024 * 1024;
     RuntimePause = class extends Error {
       outcome;
       operationId;
@@ -87742,10 +87872,15 @@ var init_worker = __esm({
         await containedPath(this.paths.home, "metadata/worker.json");
         await mkdir11(this.paths.metadata, { recursive: true });
         try {
-          this.state = JSON.parse(await readFile13(this.statePath, "utf8"));
+          const raw = await readFile13(this.statePath, "utf8");
+          this.state = JSON.parse(raw);
           if (this.state.version !== 1 || !Number.isSafeInteger(this.state.seq) || !this.state.commands || !this.state.runs) {
             throw new Error("worker checkpoint version or shape is invalid");
           }
+          if (raw.length > MAX_WORKER_STATE_CHARS) {
+            this.state = capStrings(this.state, MAX_EXTERNAL_STRING_CHARS);
+            await atomicWrite(this.statePath, JSON.stringify(this.state));
+          }
         } catch (error53) {
           if (error53.code !== "ENOENT")
             throw error53;
@@ -87974,7 +88109,7 @@ var init_worker = __esm({
             try {
               if (saved?.approvalPending && name === "call_paid_service" && economy)
                 await economy.execute("inspect_paid_service", { url: args.url, method: args.method, sample_body: args.body });
-              output3 = await invoke();
+              output3 = capToolResult(await invoke());
               if (financialState && name === "get_budget_status")
                 output3 = { ok: true, data: latestFinancialState };
               if (financialState && name === "complete_session" && output3.ok && output3.data && typeof output3.data === "object")
@@ -88417,7 +88552,7 @@ async function runAgentTaskDetailed(session, task, options = {}) {
       "You can use new services without prior CrowdCode reviews. Payable: x402 USDC on Base or Solana and MPP USDC.e on Tempo; the hosted gateway pays the cheapest offered network your balance on that network covers. Other networks, including Monad, are not payable here. Payments stay within the user's configured limits; over-limit quotes request approval. Use read_service for authenticated job polling and never resubmit a pending generation job.",
       "funding_required means your USDC is on the wrong network, not that you are out of money, and a wallet with money in it finishes the job. First go back to find_paid_service and look for an equivalent service sold on a network you already hold USDC on \u2014 every result lists its `networks`, and paying from an existing balance costs no transfer fee. If none fits, move funds with bridge_usdc and retry the same call: aim for about $1 so one fee funds many later calls, and when the wallet holds less than that, move what the call needs plus the fee, as long as the fee stays under about a fifth of what moves (roughly $0.15 and up at a $0.03 fee). A move inside that rule is ordinary work, not a decision for the owner: make it and retry, exactly as you would pay for the call itself. The spending guard applies the owner's approval mode to the fee by itself and pauses the run if it has to, so you never need permission before calling bridge_usdc. Stopping to ask whether you may spend three cents moving the owner's own money to the network a seller wants is the dead end this rule exists to prevent. Only below that threshold is a move not worth making on your own: quote_only it, then ask. When the owner asks you to move funds, quote first, state the fee, then execute and report both transaction ids and the amount received. Moves follow the owner's approval mode like payments; never move funds without a reason.",
       "When money is what blocks a task, say exactly that and nothing else: which network the service is sold on, what you hold where, and the single thing you need \u2014 permission for a move you have quoted, or USDC added on that network. Never dress a funding problem up as the service being unsuitable or unavailable, never imply your wallet is empty when it is not, and never offer a recorded request_service as though it were the answer the owner asked for."
-    ] : [], capabilityIndex ?? DEFAULT_CAPABILITY_INDEX, "When a task needs current web facts, search results, or unfamiliar page content, make one paid web search your FIRST move \u2014 do not serially guess URLs with curl; one paid search replaces minutes of blind fetching and costs less than the LLM turns it saves.", "Approval, budget, reputation, and payment rails are enforced in code \u2014 you cannot bypass them, so state costs plainly and never invent payment details.", "When the user says CrowdCode on/off, call set_crowdcode_enabled; default to session scope unless they ask for a persistent default. Use list_my_reviews and delete_my_review for user-directed review management, including while off. Before finishing a substantive task, consider concrete paid services that would fix real shortcomings and submit qualifying demand with request_service (input, output, problem, price_per_call), unless CrowdCode is off. Never invent a purchase or demand.", "Each tool result includes the budget before and after that tool call. Never ask for wallet private keys or secrets.", ...options.promptSections ?? []);
+    ] : [], capabilityIndex ?? DEFAULT_CAPABILITY_INDEX, "When a task needs current web facts, search results, or unfamiliar page content, make one paid web search your FIRST move \u2014 do not serially guess URLs with curl; one paid search replaces minutes of blind fetching and costs less than the LLM turns it saves.", "Approval, budget, reputation, and payment rails are enforced in code \u2014 you cannot bypass them, so state costs plainly and never invent payment details.", "When the user says CrowdCode on/off, call set_crowdcode_enabled; default to session scope unless they ask for a persistent default. Use list_my_reviews and delete_my_review for user-directed review management, including while off. Before finishing a substantive task, consider concrete paid services that would fix real shortcomings and submit qualifying demand with request_service (input, output, problem, price_per_call), unless CrowdCode is off. Never invent a purchase or demand.", "Generated media never reaches you as base64 you can read: when a service returns audio, an image, video or a PDF, the runtime has already written the bytes to a file and handed you {artifact_path, bytes, media_type} instead. Work from that path with run_shell \u2014 ffmpeg, python3 and node are installed, so converting a wav to the mp3 the owner asked for is one command \u2014 and leave the deliverable under artifacts/ in the format they asked for. Never pay a second time for a file you already hold.", "Each tool result includes the budget before and after that tool call. Never ask for wallet private keys or secrets.", ...options.promptSections ?? []);
   } else {
     systemPromptParts.push("Paid external services are unavailable in this run; work with local tools only and say so if the task truly requires external capability.");
   }
@@ -88694,6 +88829,7 @@ ${outputContinuationPrefix}`);
       } else {
         result = { ok: false, error: `unknown tool: ${call2.name}` };
       }
+      result = capToolResult(result);
       await refreshFinancialState();
       const budgetAfterToolCall = currentBudget();
       if (options.financialState && call2.name === "get_budget_status")
@@ -89062,6 +89198,9 @@ async function completeMockLlmCall(session, provider, messages, turn) {
     throw error53;
   }
 }
+function capToolResult(result) {
+  return result.data === void 0 ? result : { ...result, data: capStrings(result.data, MAX_EXTERNAL_STRING_CHARS) };
+}
 function toolMessagePayload(payload) {
   return truncateForModel(pruneToolPayload(payload), 12e3);
 }
@@ -91630,17 +91769,17 @@ async function writeTree(dir, tree, meta4) {
 }
 async function readTree(dir) {
   const tree = {};
-  const walk3 = async (current) => {
+  const walk4 = async (current) => {
     for (const entry of await readdir2(current, { withFileTypes: true })) {
       const path3 = join31(current, entry.name);
       if (entry.isDirectory()) {
-        await walk3(path3);
+        await walk4(path3);
       } else if (entry.name.endsWith(".md")) {
         tree[relative4(dir, path3)] = await readFile22(path3, "utf8");
       }
     }
   };
-  await walk3(dir);
+  await walk4(dir);
   let meta4;
   try {
     meta4 = JSON.parse(await readFile22(join31(dir, "meta.json"), "utf8"));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "opencrowd",
-  "version": "0.15.0",
+  "version": "0.15.2",
   "description": "A CLI agent with its own wallet — discovers x402 services, pays in USDC, and completes tasks under a local budget",
   "type": "module",
   "license": "MIT",