holycodex 0.5.1 → 0.5.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "holycodex",
-  "version": "0.5.1",
+  "version": "0.5.3",
   "description": "Lean Codex-only agent toolkit for ChatGPT Plus",
   "keywords": [
     "agents",

package/plugin/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "holycodex",
-  "version": "0.5.1",
+  "version": "0.5.3",
   "description": "Lean Codex workflows with routed skills, bounded subagents, scoped rules, Git Bash, and LSP.",
   "author": {
     "name": "David Basile Filho",

package/plugin/agents/explorer.toml

@@ -2,5 +2,5 @@ description = "Use for a narrow, read-only repository fact question with explici
 model = "gpt-5.6-luna"
 model_reasoning_effort = "low"
 developer_instructions = """
-Start: "I detect investigation intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Accept one task packet containing exact question; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject incomplete or ambiguous packets. Inspect only assigned scope. On native Windows use git_bash MCP when its run tool is available; otherwise use native shell directly without probing git_bash. Search, read, diagnose, or inspect history; never edit, install, research external facts, choose architecture, make final judgments, ask user, or broaden scope. Stop when criteria pass, evidence is unavailable, scope conflicts, or a write is required. Return exactly requested format with paths, symbols, commands, decisive output, uncertainty or blocker; no proposed extra work.
+Start: "I detect investigation intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Before any plan, skill routing, or task action, inspect the full callable tool registry, including deferred tools; treat only that registry as availability evidence. Accept one task packet containing exact question; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject incomplete or ambiguous packets. Inspect only assigned scope. On native Windows, before any shell call resolve `mcp__git_bash__run` from the full callable registry, including deferred tools, and use it when available; otherwise use native shell directly. Search, read, diagnose, or inspect history; never edit, install, research external facts, choose architecture, make final judgments, ask user, or broaden scope. Stop when criteria pass, evidence is unavailable, scope conflicts, or a write is required. Return exactly requested format with paths, symbols, commands, decisive output, uncertainty or blocker; no proposed extra work.
 """

package/plugin/agents/librarian.toml

@@ -2,5 +2,5 @@ description = "Use for a narrow current external-fact question requiring primary
 model = "gpt-5.6-luna"
 model_reasoning_effort = "low"
 developer_instructions = """
-Start: "I detect research intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Accept one task packet containing exact question; repository root only as context; allowed paths and symbols, normally none; forbidden paths and sources; allowed external sources; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject incomplete or ambiguous packets. On native Windows use git_bash MCP when its run tool is available; otherwise use native shell directly without probing git_bash. Prefer current primary sources; distinguish source fact from inference. Never inspect repo beyond supplied context, implement, write externally, choose architecture, make final judgments, ask user, or broaden scope. Stop when criteria pass, authoritative evidence is unavailable or conflicting, or answer needs repository work. Return exactly requested format with claim-linked source, version or date, conflicts, inference, uncertainty or blocker; no proposed extra work.
+Start: "I detect research intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Before any plan, skill routing, or task action, inspect the full callable tool registry, including deferred tools; treat only that registry as availability evidence. Accept one task packet containing exact question; repository root only as context; allowed paths and symbols, normally none; forbidden paths and sources; allowed external sources; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject incomplete or ambiguous packets. On native Windows, before any shell call resolve `mcp__git_bash__run` from the full callable registry, including deferred tools, and use it when available; otherwise use native shell directly. Prefer current primary sources; distinguish source fact from inference. Never inspect repo beyond supplied context, implement, write externally, choose architecture, make final judgments, ask user, or broaden scope. Stop when criteria pass, authoritative evidence is unavailable or conflicting, or answer needs repository work. Return exactly requested format with claim-linked source, version or date, conflicts, inference, uncertainty or blocker; no proposed extra work.
 """

package/plugin/agents/worker.toml

@@ -2,5 +2,5 @@ description = "Use for an isolated implementation slice with fixed files and acc
 model = "gpt-5.6-luna"
 model_reasoning_effort = "medium"
 developer_instructions = """
-Start: "I detect implementation intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Accept one task packet containing exact implementation outcome; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject or stop on ambiguity, overlap, architecture choice, coupled out-of-scope work, user-work conflict, or missing authority. On native Windows use git_bash MCP when its run tool is available; otherwise use native shell directly without probing git_bash. Prompt, skill, or instruction task: load caveman first; preserve constraints. Load required skills before governed action. Change only assigned files. Preserve architecture, API, behavior, naming, style, and user work. Add no speculative cleanup, abstraction, dependency, docs, or scope. Run specified proof; never claim integration or final verification. Return exactly requested format with changed paths and behavior, command results, criteria status, blocker or residual risk; no proposed extra work.
+Start: "I detect implementation intent — [reason]. [action]." Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Before any plan, skill routing, or task action, inspect the full callable tool registry, including deferred tools; treat only that registry as availability evidence. Accept one task packet containing exact implementation outcome; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Reject or stop on ambiguity, overlap, architecture choice, coupled out-of-scope work, user-work conflict, or missing authority. On native Windows, before any shell call resolve `mcp__git_bash__run` from the full callable registry, including deferred tools, and use it when available; otherwise use native shell directly. Prompt, skill, or instruction task: load caveman first; preserve constraints. Load required skills before governed action. Change only assigned files. Preserve architecture, API, behavior, naming, style, and user work. Add no speculative cleanup, abstraction, dependency, docs, or scope. Run specified proof; never claim integration or final verification. Return exactly requested format with changed paths and behavior, command results, criteria status, blocker or residual risk; no proposed extra work.
 """

package/plugin/runtime/cli.js

@@ -122,7 +122,7 @@ function paths(home = process.env.CODEX_HOME ?? join(homedir(), ".codex")) {
 		config: join(home, "config.toml"),
 		marketplaceCache,
 		cacheRoot,
-		cache: join(cacheRoot, "0.5.1"),
+		cache: join(cacheRoot, "0.5.3"),
 		agents: join(home, "holycodex", "agents"),
 		legacy: [
 			join(home, "plugins", "cache", "sisyphuslabs", "omo"),
@@ -210,7 +210,7 @@ async function cleanup(_options) {
 }
 //#endregion
 //#region src/cli.ts
-var VERSION = "0.5.1";
+var VERSION = "0.5.3";
 var HELP = `HolyCodex ${VERSION}\n\nUsage: holycodex <install|cleanup> [options]\n\nOptions:\n  --help                 Show help\n  --version              Show version\n  --no-tui               Accepted; commands are noninteractive\n  --codex-autonomous     Set autonomous Codex permissions (default)\n  --no-codex-autonomous  Preserve existing Codex permissions\n  --json                 Print machine-readable result\n`;
 async function main() {
 	const args = process$1.argv.slice(2);

package/plugin/runtime/core-instructions.js

@@ -1,4 +1,4 @@
 //#region src/core-instructions.ts
-var CORE_INSTRUCTIONS = "HolyCodex: Start first user-facing update: \"I detect [fix/implementation/investigation/question] intent — [reason]. [action].\" Act. Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Shell: on native Windows, use git_bash MCP for shell commands when its run tool is available; otherwise use the native shell directly without probing git_bash. Prompt, skill, or instruction edit: load caveman; preserve constraints. Main agent owns intent, scope, architecture, decisions, integration, and final verification. Load each applicable skill before its governed action; a skill alone is sufficient when the main agent can complete the work directly. Delegate only a bounded, independent, unambiguous slice cheaper to integrate than local work and suitable for its Luna model: explorer=read-only exact repo facts; librarian=read-only current external facts from primary sources; worker=isolated fixed-scope implementation. Never delegate trivial or coupled work, architecture, user clarification, integration, final judgment, or final verification. Every packet states exact question or outcome; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Use concrete names, paths, APIs, functions, tests, versions, and expected observations when known; include no unrelated transcript. Treat returns as input: inspect, reconcile, integrate, and verify. Stop when packet is satisfied or blocked; never recurse or create agent organizations. Use GPT 5.6 Luna low for explorer and librarian; use GPT 5.6 Luna medium for worker.";
+var CORE_INSTRUCTIONS = "HolyCodex: Start first user-facing update: \"I detect [fix/implementation/investigation/question] intent — [reason]. [action].\" Act. Default user-facing replies: grammatical sentences; no filler or hedging. Preserve technical terms, code, paths, error text, and commit keywords; use full grammar for safety warnings, irreversible confirmations, ordered steps, ambiguity, or clarification. Before any plan, skill routing, delegation, or task action, inspect the full callable tool registry, including deferred tools; treat only that registry as availability evidence. Shell: on native Windows, before any shell call resolve `mcp__git_bash__run` from the full callable registry, including deferred tools, and use it when available; otherwise use the native shell directly. Prompt, skill, or instruction edit: load caveman; preserve constraints. Main agent owns intent, scope, architecture, decisions, integration, and final verification. Load each applicable skill before its governed action; a skill alone is sufficient when the main agent can complete the work directly. Delegate only a bounded, independent, unambiguous slice cheaper to integrate than local work and suitable for its Luna model: explorer=read-only exact repo facts; librarian=read-only current external facts from primary sources; worker=isolated fixed-scope implementation. Never delegate trivial or coupled work, architecture, user clarification, integration, final judgment, or final verification. Every packet states exact question or outcome; repository root; allowed paths and symbols; forbidden paths; relevant architecture and existing behavior; required skills; exact inputs; output format; acceptance criteria; required commands or evidence; unchanged constraints; prohibited expansion; known uncertainty; blocker behavior; exact stop condition. Use concrete names, paths, APIs, functions, tests, versions, and expected observations when known; include no unrelated transcript. Treat returns as input: inspect, reconcile, integrate, and verify. Stop when packet is satisfied or blocked; never recurse or create agent organizations. Use GPT 5.6 Luna low for explorer and librarian; use GPT 5.6 Luna medium for worker.";
 //#endregion
 export { CORE_INSTRUCTIONS as t };

package/plugin/runtime/git-bash.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { a as successResponse, n as errorResponse, o as isPlainRecord, r as jsonRpcId, t as runJsonRpcStdioServer } from "./mcp-stdio-core.js";
+import { a as stackOrMessageFromError, i as messageFromError, n as errorResponse, o as successResponse, r as jsonRpcId, s as isPlainRecord, t as runJsonRpcStdioServer } from "./mcp-stdio-core.js";
 import { join } from "node:path";
 import { argv, stderr } from "node:process";
 import { tmpdir } from "node:os";
@@ -166,7 +166,7 @@ async function handleGitBashMcpRequest(input, options = {}) {
 		capabilities: { tools: { listChanged: false } },
 		serverInfo: {
 			name: "git_bash",
-			version: "0.5.1"
+			version: "0.5.3"
 		},
 		protocolVersion: protocolVersionFromInput(input) ?? "2024-11-05"
 	});
@@ -216,7 +216,7 @@ async function runToolResponse(id, args, options) {
 		});
 		return toolResponse(id, JSON.stringify(result, null, 2));
 	} catch (error) {
-		return toolResponse(id, error instanceof Error ? error.message : String(error), true);
+		return toolResponse(id, messageFromError(error), true);
 	}
 }
 function toolsForOptions(options) {
@@ -349,7 +349,7 @@ async function main() {
 	process.exitCode = 2;
 }
 main().catch((error) => {
-	stderr.write(`${error instanceof Error ? error.stack ?? error.message : String(error)}\n`);
+	stderr.write(`${stackOrMessageFromError(error)}\n`);
 	process.exitCode = 1;
 });
 //#endregion

package/plugin/runtime/lsp.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { a as successResponse, i as messageFromError, n as errorResponse, o as isPlainRecord, r as jsonRpcId, t as runJsonRpcStdioServer } from "./mcp-stdio-core.js";
+import { a as stackOrMessageFromError, i as messageFromError, n as errorResponse, o as successResponse, r as jsonRpcId, s as isPlainRecord, t as runJsonRpcStdioServer } from "./mcp-stdio-core.js";
 import { basename, delimiter, dirname, extname, isAbsolute, join, relative, resolve } from "node:path";
 import { argv, execPath, stderr } from "node:process";
 import { homedir, tmpdir } from "node:os";
@@ -94,7 +94,7 @@ function isLspDeadConnectionError(err) {
 //#region packages/lsp-core/src/lsp/cleanup-errors.ts
 function reportBestEffortCleanupError(operation, error) {
 	if (process.env["CODEX_LSP_DEBUG_CLEANUP"] !== "1") return;
-	const message = error instanceof Error ? error.message : String(error);
+	const message = messageFromError(error);
 	console.error(`[codex-lsp] ignored ${operation} failure during cleanup: ${message}`);
 }
 var STOP_HARD_KILL_TIMEOUT_MS = 5e3;
@@ -328,15 +328,12 @@ function parseContentLength(headers) {
 	}
 	return null;
 }
-function isJsonRpcObject(value) {
-	return typeof value === "object" && value !== null && !Array.isArray(value);
-}
 function getMessageId(message) {
 	const id = message["id"];
 	if (typeof id === "number" || typeof id === "string" || id === null) return id;
 }
 function jsonRpcErrorToError(value) {
-	if (!isJsonRpcObject(value)) return /* @__PURE__ */ new Error("JSON-RPC request failed");
+	if (!isPlainRecord(value)) return /* @__PURE__ */ new Error("JSON-RPC request failed");
 	const message = typeof value["message"] === "string" ? value["message"] : "JSON-RPC request failed";
 	const error = new Error(message);
 	if (typeof value["code"] === "number") error.name = `JsonRpcError(${value["code"]})`;
@@ -499,21 +496,18 @@ function spawnProcess(command, options) {
 }
 //#endregion
 //#region packages/lsp-core/src/lsp/transport.ts
-function isRecord$3(value) {
-	return typeof value === "object" && value !== null && !Array.isArray(value);
-}
 function parseConfigurationItems(params) {
-	if (!isRecord$3(params) || !Array.isArray(params["items"])) return [];
+	if (!isPlainRecord(params) || !Array.isArray(params["items"])) return [];
 	const items = [];
 	for (const item of params["items"]) {
-		if (!isRecord$3(item)) continue;
+		if (!isPlainRecord(item)) continue;
 		const section = item["section"];
 		items.push(section === void 0 || typeof section !== "string" ? {} : { section });
 	}
 	return items;
 }
 function parseDiagnosticsParams(params) {
-	if (!isRecord$3(params) || typeof params["uri"] !== "string") return null;
+	if (!isPlainRecord(params) || typeof params["uri"] !== "string") return null;
 	const diagnostics = Array.isArray(params["diagnostics"]) ? params["diagnostics"].filter(isDiagnostic) : [];
 	return {
 		uri: params["uri"],
@@ -680,13 +674,13 @@ function createLspSpawnEnv(_root, input) {
 	return { ...input };
 }
 function isDiagnostic(value) {
-	return isRecord$3(value) && isRange(value["range"]) && typeof value["message"] === "string";
+	return isPlainRecord(value) && isRange(value["range"]) && typeof value["message"] === "string";
 }
 function isRange(value) {
-	return isRecord$3(value) && isPosition(value["start"]) && isPosition(value["end"]);
+	return isPlainRecord(value) && isPosition(value["start"]) && isPosition(value["end"]);
 }
 function isPosition(value) {
-	return isRecord$3(value) && typeof value["line"] === "number" && typeof value["character"] === "number";
+	return isPlainRecord(value) && typeof value["line"] === "number" && typeof value["character"] === "number";
 }
 //#endregion
 //#region packages/lsp-core/src/lsp/connection.ts
@@ -969,10 +963,9 @@ var LspClient = class extends LspClientConnection {
 		});
 	}
 	async definition(filePath, line, character) {
-		const absPath = resolve(contextCwd(), filePath);
-		await this.openFile(absPath);
+		const textDocument = await this.openTextDocument(filePath);
 		return this.sendRequest("textDocument/definition", {
-			textDocument: { uri: pathToFileURL(absPath).href },
+			textDocument,
 			position: {
 				line: line - 1,
 				character
@@ -980,10 +973,9 @@ var LspClient = class extends LspClientConnection {
 		});
 	}
 	async references(filePath, line, character, includeDeclaration = true) {
-		const absPath = resolve(contextCwd(), filePath);
-		await this.openFile(absPath);
+		const textDocument = await this.openTextDocument(filePath);
 		return this.sendRequest("textDocument/references", {
-			textDocument: { uri: pathToFileURL(absPath).href },
+			textDocument,
 			position: {
 				line: line - 1,
 				character
@@ -992,9 +984,8 @@ var LspClient = class extends LspClientConnection {
 		});
 	}
 	async documentSymbols(filePath) {
-		const absPath = resolve(contextCwd(), filePath);
-		await this.openFile(absPath);
-		return this.sendRequest("textDocument/documentSymbol", { textDocument: { uri: pathToFileURL(absPath).href } });
+		const textDocument = await this.openTextDocument(filePath);
+		return this.sendRequest("textDocument/documentSymbol", { textDocument });
 	}
 	async workspaceSymbols(query) {
 		return this.sendRequest("workspace/symbol", { query });
@@ -1005,9 +996,7 @@ var LspClient = class extends LspClientConnection {
 		return /unsupported|not supported|method not found|unknown request/i.test(error.message);
 	}
 	async diagnostics(filePath) {
-		const absPath = resolve(contextCwd(), filePath);
-		const uri = pathToFileURL(absPath).href;
-		await this.openFile(absPath);
+		const { uri } = await this.openTextDocument(filePath);
 		await new Promise((r) => setTimeout(r, POST_DIAGNOSTICS_WAIT_MS));
 		try {
 			const result = await this.sendRequest("textDocument/diagnostic", { textDocument: { uri } });
@@ -1018,10 +1007,9 @@ var LspClient = class extends LspClientConnection {
 		return { items: this.getStoredDiagnostics(uri) };
 	}
 	async prepareRename(filePath, line, character) {
-		const absPath = resolve(contextCwd(), filePath);
-		await this.openFile(absPath);
+		const textDocument = await this.openTextDocument(filePath);
 		return this.sendRequest("textDocument/prepareRename", {
-			textDocument: { uri: pathToFileURL(absPath).href },
+			textDocument,
 			position: {
 				line: line - 1,
 				character
@@ -1029,10 +1017,9 @@ var LspClient = class extends LspClientConnection {
 		});
 	}
 	async rename(filePath, line, character, newName) {
-		const absPath = resolve(contextCwd(), filePath);
-		await this.openFile(absPath);
+		const textDocument = await this.openTextDocument(filePath);
 		return this.sendRequest("textDocument/rename", {
-			textDocument: { uri: pathToFileURL(absPath).href },
+			textDocument,
 			position: {
 				line: line - 1,
 				character
@@ -1040,6 +1027,11 @@ var LspClient = class extends LspClientConnection {
 			newName
 		});
 	}
+	async openTextDocument(filePath) {
+		const absPath = resolve(contextCwd(), filePath);
+		await this.openFile(absPath);
+		return { uri: pathToFileURL(absPath).href };
+	}
 };
 //#endregion
 //#region packages/lsp-core/src/lsp/process-signal-cleanup.ts
@@ -1352,15 +1344,12 @@ function writeInstallDecisions(decisions) {
 	renameSync(tmpPath, path);
 }
 function isInstallDecisions(value) {
-	return isRecord$2(value) && Object.values(value).every(isInstallDecisionRecord);
+	return isPlainRecord(value) && Object.values(value).every(isInstallDecisionRecord);
 }
 function isInstallDecisionRecord(value) {
-	if (!isRecord$2(value)) return false;
+	if (!isPlainRecord(value)) return false;
 	return isInstallDecision(value["decision"]) && typeof value["decidedAt"] === "string";
 }
-function isRecord$2(value) {
-	return typeof value === "object" && value !== null && !Array.isArray(value);
-}
 //#endregion
 //#region packages/lsp-core/src/lsp/server-definitions.ts
 var LSP_INSTALL_HINTS = {
@@ -1826,31 +1815,28 @@ function applyOptionalServerFields(server, entry) {
 	if (entry.initialization !== void 0) server.initialization = entry.initialization;
 }
 function isConfigJson(value) {
-	if (!isRecord$1(value)) return false;
+	if (!isPlainRecord(value)) return false;
 	const lsp = value["lsp"];
-	return lsp === void 0 || isRecord$1(lsp);
+	return lsp === void 0 || isPlainRecord(lsp);
 }
 function parseLspEntry(value) {
 	return isLspEntry(value) ? value : null;
 }
 function isLspEntry(value) {
-	if (!isRecord$1(value)) return false;
+	if (!isPlainRecord(value)) return false;
 	const disabled = value["disabled"];
 	const command = value["command"];
 	const extensions = value["extensions"];
 	const priority = value["priority"];
 	const env = value["env"];
 	const initialization = value["initialization"];
-	return (disabled === void 0 || typeof disabled === "boolean") && (command === void 0 || isStringArray(command)) && (extensions === void 0 || isStringArray(extensions)) && (priority === void 0 || typeof priority === "number") && (env === void 0 || isStringRecord$1(env)) && (initialization === void 0 || isRecord$1(initialization));
+	return (disabled === void 0 || typeof disabled === "boolean") && (command === void 0 || isStringArray(command)) && (extensions === void 0 || isStringArray(extensions)) && (priority === void 0 || typeof priority === "number") && (env === void 0 || isStringRecord$1(env)) && (initialization === void 0 || isPlainRecord(initialization));
 }
 function isStringArray(value) {
 	return Array.isArray(value) && value.every((item) => typeof item === "string");
 }
 function isStringRecord$1(value) {
-	return isRecord$1(value) && Object.values(value).every((item) => typeof item === "string");
-}
-function isRecord$1(value) {
-	return typeof value === "object" && value !== null && !Array.isArray(value);
+	return isPlainRecord(value) && Object.values(value).every((item) => typeof item === "string");
 }
 function getDisabledServerIds() {
 	const configs = loadAllConfigs();
@@ -2311,9 +2297,7 @@ var RUST_SRC_REPAIR_MESSAGE = [
 	"  rustup component remove rust-src",
 	"  rustup component add rust-src"
 ];
-function errorMessage$1(error) {
-	return error instanceof Error ? error.message : String(error);
-}
+var errorMessage = messageFromError;
 function formatKnownLspStartupFailure(error) {
 	if (!(error instanceof LspProcessExitedError)) return null;
 	if (error.serverId !== "rust") return null;
@@ -2330,7 +2314,7 @@ function formatKnownLspStartupFailure(error) {
 function handleMissingDependencyError(error) {
 	const knownStartupFailure = formatKnownLspStartupFailure(error);
 	if (knownStartupFailure) return knownStartupFailure;
-	const message = errorMessage$1(error);
+	const message = errorMessage(error);
 	return message.includes("NOT INSTALLED") || message.includes("No LSP server configured") ? message : null;
 }
 //#endregion
@@ -2350,10 +2334,20 @@ function missingDependencyResult(error, details) {
 		}
 	};
 }
+function missingDependencyResultOrThrow(error, details) {
+	const result = missingDependencyResult(error, details);
+	if (result !== null) return result;
+	throw error;
+}
 //#endregion
 //#region packages/lsp-core/src/tools/parameters.ts
-function isRecord(value) {
-	return typeof value === "object" && value !== null && !Array.isArray(value);
+var isRecord = isPlainRecord;
+function sourcePosition(params) {
+	return {
+		filePath: requireString(params, "filePath"),
+		line: requireNumber(params, "line"),
+		character: requireNumber(params, "character")
+	};
 }
 function requireString(params, key) {
 	const value = params[key];
@@ -2449,7 +2443,7 @@ async function executeLspDiagnostics(params, signal) {
 			truncated
 		});
 	} catch (error) {
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			severity,
 			mode: "file",
@@ -2457,8 +2451,6 @@ async function executeLspDiagnostics(params, signal) {
 			totalDiagnostics: 0,
 			truncated: false
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 //#endregion
@@ -2484,9 +2476,7 @@ async function executeLspInstallDecision(params) {
 //#endregion
 //#region packages/lsp-core/src/tools/navigation.ts
 async function executeLspGotoDefinition(params, signal) {
-	const filePath = requireString(params, "filePath");
-	const line = requireNumber(params, "line");
-	const character = requireNumber(params, "character");
+	const { filePath, line, character } = sourcePosition(params);
 	try {
 		const result = await withLspClient(filePath, async (client) => client.definition(filePath, line, character), "definition", clientOptions(signal));
 		const locations = !result ? [] : Array.isArray(result) ? result : [result];
@@ -2499,20 +2489,16 @@ async function executeLspGotoDefinition(params, signal) {
 		if (locations.length === 0) return text("No definition found", details);
 		return text(locations.map(formatLocation).join("\n"), details);
 	} catch (error) {
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			line,
 			character,
 			locations: []
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 async function executeLspFindReferences(params, signal) {
-	const filePath = requireString(params, "filePath");
-	const line = requireNumber(params, "line");
-	const character = requireNumber(params, "character");
+	const { filePath, line, character } = sourcePosition(params);
 	const includeDeclaration = optionalBoolean(params, "includeDeclaration") ?? true;
 	try {
 		const result = await withLspClient(filePath, async (client) => client.references(filePath, line, character, includeDeclaration), "references", clientOptions(signal));
@@ -2531,7 +2517,7 @@ async function executeLspFindReferences(params, signal) {
 		if (total === 0) return text("No references found", details);
 		return text([...truncated ? [`Found ${total} references (showing first 200):`] : [], ...limited.map(formatLocation)].join("\n"), details);
 	} catch (error) {
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			line,
 			character,
@@ -2539,15 +2525,10 @@ async function executeLspFindReferences(params, signal) {
 			totalReferences: 0,
 			truncated: false
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 //#endregion
 //#region packages/lsp-core/src/lsp/workspace-edit.ts
-function errorMessage(error) {
-	return error instanceof Error ? error.message : String(error);
-}
 function isPathInsideWorkspace(filePath, workspaceRoot) {
 	const relativePath = relative(workspaceRoot, filePath);
 	return relativePath === "" || !relativePath.startsWith("..") && !isAbsolute(relativePath);
@@ -2564,7 +2545,7 @@ function uriToWorkspacePath(uri, workspaceRoot) {
 	} catch (error) {
 		return {
 			success: false,
-			error: `non-file URI ${uri}: ${errorMessage(error)}`
+			error: `non-file URI ${uri}: ${messageFromError(error)}`
 		};
 	}
 	let validatedPath;
@@ -2573,7 +2554,7 @@ function uriToWorkspacePath(uri, workspaceRoot) {
 	} catch (error) {
 		return {
 			success: false,
-			error: `${filePath}: ${errorMessage(error)}`
+			error: `${filePath}: ${messageFromError(error)}`
 		};
 	}
 	if (!isPathInsideWorkspace(validatedPath, workspaceRoot)) return {
@@ -2717,9 +2698,7 @@ function applyWorkspaceEdit(edit, options = {}) {
 //#endregion
 //#region packages/lsp-core/src/tools/rename.ts
 async function executeLspPrepareRename(params, signal) {
-	const filePath = requireString(params, "filePath");
-	const line = requireNumber(params, "line");
-	const character = requireNumber(params, "character");
+	const { filePath, line, character } = sourcePosition(params);
 	try {
 		const result = await withLspClient(filePath, async (client) => client.prepareRename(filePath, line, character), "prepareRename", clientOptions(signal));
 		const details = {
@@ -2730,20 +2709,16 @@ async function executeLspPrepareRename(params, signal) {
 		};
 		return text(formatPrepareRenameResult(result), details);
 	} catch (error) {
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			line,
 			character,
 			result: null
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 async function executeLspRename(params, signal) {
-	const filePath = requireString(params, "filePath");
-	const line = requireNumber(params, "line");
-	const character = requireNumber(params, "character");
+	const { filePath, line, character } = sourcePosition(params);
 	const newName = requireString(params, "newName");
 	try {
 		const edit = await withLspClient(filePath, async (client, workspaceRoot) => ({
@@ -2761,7 +2736,7 @@ async function executeLspRename(params, signal) {
 		};
 		return text(formatApplyResult(apply), details, !apply.success);
 	} catch (error) {
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			line,
 			character,
@@ -2769,8 +2744,6 @@ async function executeLspRename(params, signal) {
 			apply: null,
 			edit: null
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 //#endregion
@@ -2838,7 +2811,7 @@ async function executeLspSymbols(params, signal) {
 		return formatSymbolsResult(filePath, scope, await withLspClient(filePath, async (client) => client.documentSymbols(filePath), "documentSymbols", clientOptions(signal)), limit);
 	} catch (error) {
 		const query = optionalString(params, "query");
-		const missingDependency = missingDependencyResult(error, {
+		return missingDependencyResultOrThrow(error, {
 			filePath,
 			scope,
 			symbols: [],
@@ -2846,8 +2819,6 @@ async function executeLspSymbols(params, signal) {
 			truncated: false,
 			...query === void 0 ? {} : { query }
 		});
-		if (missingDependency) return missingDependency;
-		throw error;
 	}
 }
 function formatSymbolsResult(filePath, scope, symbols, limit, query) {
@@ -2872,6 +2843,21 @@ function formatSymbolsResult(filePath, scope, symbols, limit, query) {
 }
 //#endregion
 //#region packages/lsp-core/src/tools/definitions.ts
+var POSITION_PROPERTIES = {
+	line: {
+		type: "number",
+		description: "1-based line number."
+	},
+	character: {
+		type: "number",
+		description: "0-based column."
+	}
+};
+var POSITION_REQUIRED = [
+	"filePath",
+	"line",
+	"character"
+];
 var LSP_MCP_TOOLS = [
 	{
 		name: "status",
@@ -2915,19 +2901,8 @@ var LSP_MCP_TOOLS = [
 				type: "string",
 				description: "Source file containing the symbol."
 			},
-			line: {
-				type: "number",
-				description: "1-based line number."
-			},
-			character: {
-				type: "number",
-				description: "0-based column."
-			}
-		}, [
-			"filePath",
-			"line",
-			"character"
-		]),
+			...POSITION_PROPERTIES
+		}, POSITION_REQUIRED),
 		execute: executeLspGotoDefinition
 	},
 	{
@@ -2940,23 +2915,12 @@ var LSP_MCP_TOOLS = [
 				type: "string",
 				description: "Source file containing the symbol."
 			},
-			line: {
-				type: "number",
-				description: "1-based line number."
-			},
-			character: {
-				type: "number",
-				description: "0-based column."
-			},
+			...POSITION_PROPERTIES,
 			includeDeclaration: {
 				type: "boolean",
 				description: "Include the declaration. Defaults to true."
 			}
-		}, [
-			"filePath",
-			"line",
-			"character"
-		]),
+		}, POSITION_REQUIRED),
 		execute: executeLspFindReferences
 	},
 	{
@@ -2995,19 +2959,8 @@ var LSP_MCP_TOOLS = [
 				type: "string",
 				description: "Source file path."
 			},
-			line: {
-				type: "number",
-				description: "1-based line number."
-			},
-			character: {
-				type: "number",
-				description: "0-based column."
-			}
-		}, [
-			"filePath",
-			"line",
-			"character"
-		]),
+			...POSITION_PROPERTIES
+		}, POSITION_REQUIRED),
 		execute: executeLspPrepareRename
 	},
 	{
@@ -3020,14 +2973,7 @@ var LSP_MCP_TOOLS = [
 				type: "string",
 				description: "Source file path."
 			},
-			line: {
-				type: "number",
-				description: "1-based line number."
-			},
-			character: {
-				type: "number",
-				description: "0-based column."
-			},
+			...POSITION_PROPERTIES,
 			newName: {
 				type: "string",
 				description: "New symbol name."
@@ -3072,7 +3018,7 @@ function coerceToolArguments(value) {
 //#endregion
 //#region packages/lsp-core/src/mcp.ts
 var SERVER_NAME = "lsp";
-var SERVER_VERSION = "0.5.1";
+var SERVER_VERSION = "0.5.3";
 async function handleLspMcpRequest(input) {
 	if (!isPlainRecord(input)) return errorResponse(null, -32600, "Invalid Request");
 	const id = jsonRpcId(input["id"]);
@@ -3439,7 +3385,7 @@ function daemonUnreachableResult(paths, error) {
 		content: [{
 			type: "text",
 			text: [
-				`LSP daemon unreachable: ${error instanceof Error ? error.message : String(error)}.`,
+				`LSP daemon unreachable: ${messageFromError(error)}.`,
 				"The MCP server is a thin proxy and never runs language servers in-process.",
 				`Socket: ${paths.socket}`,
 				`Logs: ${paths.log}`,
@@ -3509,7 +3455,7 @@ async function runMcpStdioProxy(options = {}) {
 			...options.ensure ? { ensure: options.ensure } : {}
 		},
 		onHandlerError: (error) => {
-			process.stderr.write(`[lsp-daemon] proxy error: ${error instanceof Error ? error.message : String(error)}\n`);
+			process.stderr.write(`[lsp-daemon] proxy error: ${messageFromError(error)}\n`);
 		}
 	});
 }
@@ -3659,7 +3605,7 @@ async function main() {
 	process.exitCode = 2;
 }
 main().catch((error) => {
-	stderr.write(`${error instanceof Error ? error.stack ?? error.message : String(error)}\n`);
+	stderr.write(`${stackOrMessageFromError(error)}\n`);
 	process.exitCode = 1;
 });
 //#endregion

package/plugin/runtime/mcp-stdio-core.js

@@ -31,6 +31,9 @@ function jsonRpcId(value) {
 function messageFromError(error) {
 	return error instanceof Error ? error.message : String(error);
 }
+function stackOrMessageFromError(error) {
+	return error instanceof Error ? error.stack ?? error.message : String(error);
+}
 //#endregion
 //#region packages/mcp-stdio-core/src/transport.ts
 var HEADER_SEPARATOR = Buffer.from("\r\n\r\n");
@@ -118,7 +121,7 @@ function parseJsonPayload(payload, responseMode) {
 	} catch (error) {
 		return {
 			kind: "parse_error",
-			message: error instanceof Error ? error.message : String(error),
+			message: messageFromError(error),
 			responseMode
 		};
 	}
@@ -206,4 +209,4 @@ function createIdleTimer(idleTimeoutMs, log, onIdleTimeout) {
 	};
 }
 //#endregion
-export { successResponse as a, messageFromError as i, errorResponse as n, isPlainRecord as o, jsonRpcId as r, runJsonRpcStdioServer as t };
+export { stackOrMessageFromError as a, messageFromError as i, errorResponse as n, successResponse as o, jsonRpcId as r, isPlainRecord as s, runJsonRpcStdioServer as t };

package/plugin/skills/define-goal/SKILL.md

@@ -13,15 +13,15 @@ Create one bounded, verifiable objective; no plan, log, snapshot, ledger, handof
 
 ## Flow
 
-1. State concrete desired outcome, target, completion criteria, acceptable proof, scope, exclusions, input blocker, and explicit stop condition.
+1. State concrete outcome, target, completion criteria, proof, scope, exclusions, input blocker, and stop condition.
 2. Add useful checks, paths, environments, counts, or limits; no fake precision or quality ratchet.
-3. Replace activity goals like “improve” with observable state. Bound subjective quality by the approved task and named validator.
+3. Replace activity goals such as “improve” with observable state. Bound subjective quality by approved task and named validator.
 4. Ask one short question only if missing scope or proof would materially change intent.
 5. Call `get_goal`.
    - No goal: quality-check, then `create_goal`.
    - Same active goal: reuse.
    - Conflicting goal: ask whether to finish it or use another task.
-6. Call `create_goal` with one concise objective. Set token budget only when user explicitly asks. Stop once the objective's criteria pass; do not extend it for polishing, speculative expansion, repeated review, adjacent work, or newly noticed opportunities.
+6. Call `create_goal` with one concise objective. Set token budget only when explicitly requested. Stop when criteria pass; do not extend for polish, speculation, repeated review, adjacent work, or new opportunities.
 
 ## Quality gate
 

package/plugin/skills/frontend/SKILL.md

@@ -5,46 +5,46 @@ description: Use when a task needs frontend, web UI, UX, visual design, interact
 
 # Frontend
 
-Preserve stack, behavior, tokens, patterns, supplied visual contract, and product identity. Keep scope small. Avoid generic AI styling. Do not force `DESIGN.md`, research, dependencies, GSAP, or heavy QA.
+Preserve stack, behavior, tokens, patterns, visual contract, and identity. Keep scope small; avoid generic AI styling. Do not force `DESIGN.md`, research, dependencies, GSAP, or heavy QA.
 
 ## Approval sequence
 
 1. Load `frontend`; inspect the request, product shell, target surface, design tokens, nearest pattern, responsive contract, supplied references, and relevant states.
-2. Select only material design decisions: visual direction, typography, layout and density, color and surfaces, assets, responsive behavior, applicable states, plus a motion system and accessibility treatment for every task. Define how motion communicates hierarchy, state, interaction, continuity, or navigation; keep it proportional. Define `prefers-reduced-motion`, keyboard operation, focus visibility, semantics, contrast, and labels. Preserve established decisions unless change is requested.
+2. Select only material decisions: direction, type, layout/density, color/surfaces, assets, responsiveness, applicable states, plus a motion system and accessibility treatment for every task. Give motion a proportional hierarchy, state, interaction, continuity, or navigation purpose. Define `prefers-reduced-motion`, keyboard operation, focus visibility, semantics, contrast, and labels. Preserve established decisions unless change is requested.
 3. Present the compact decision set with existing constraints and ask for approval before implementation. For a fix, audit, accessibility, or performance task, include only decisions that alter user-visible design; diagnosis and technical implementation details need no approval.
 4. After approval, ask whether the user wants to define a goal. Only after explicit agreement load `define-goal`; otherwise implement.
 
-Approval owns visible direction and material interaction changes. Existing component APIs, code structure, exact CSS values, breakpoint mechanics, libraries already required by the approved route, and other reversible implementation details remain implementation decisions.
+Approval owns visible direction and material interaction. Component APIs, code structure, exact CSS, breakpoint mechanics, route-required libraries, and other reversible implementation details remain implementation decisions.
 
 ## Route
 
-- Build/redesign: inspect shell, tokens, target component, responsive contract.
-- Visual reference: use it first; measure layout, type, color, spacing, surface, and motion.
+- Build/redesign: inspect shell, tokens, component, responsive contract.
+- Visual reference: use first; measure layout, type, color, spacing, surface, motion.
 - Performance: load `references/perfection/README.md` only for real audit or regression.
 - Palette/type/style: load only relevant `references/ui-ux-db` data.
 - Motion-rich premium React/Tailwind page creation or redesign: mandatory GPT Taste route below. Ordinary fixes, debugging, accessibility, and performance work stay on normal route.
 
 ## GPT Taste route
 
-- During approval, emit `<design_plan>` with prompt-derived visual direction, type, layout, motion, responsive, accessibility, and state choices. Established contracts win.
+- During approval, emit `<design_plan>` with prompt-derived direction, type, layout, motion, responsive, accessibility, and state choices. Established contracts win.
 - Treat AIDA, bento layouts, font shortlists, GSAP, Picsum, component counts, and named motion patterns as options only when they fit product, task, and stack.
-- For each chosen pattern, state its user-facing purpose and containment. Do not add dependencies or remote assets unless approved and required.
+- State each chosen pattern's user-facing purpose and containment. Add no dependency or remote asset unless approved and required.
 - Motion is required but may be subtle. Avoid gratuitous, blocking, performance-heavy, or reduced-motion-unsafe effects.
 - Prevent horizontal overflow, clipped content, dead layouts, empty cards, fake dashboards, copied text, and inaccessible interaction.
 
 Ban emojis; cheap/generic meta-labels (`SECTION 01`, `QUESTION 05`, `ABOUT US`); invisible button text; empty bento cells/cards; narrow multi-line/centered heroes; repeated left/right or flat sections/backgrounds; stock-feeling imagery; fake dashboards; meaningless gradients; copied reference text.
 
-Required `<design_plan>`: visible direction; layout and responsive contract; asset role; motion purpose and reduced-motion fallback; accessibility treatment; applicable loading, error, and empty states. Include only chosen patterns, not a fixed architecture count.
+Required `<design_plan>`: visible direction; layout/responsive contract; asset role; motion purpose/reduced-motion fallback; accessibility treatment; applicable loading, error, and empty states. Include only chosen patterns, never a fixed architecture count.
 
 Only after approval and optional goal choice may implementation begin.
 
 ## Implementation
 
-1. Inspect target and nearest established pattern.
-2. Lock visual direction: type, spacing, surfaces, color, asset role, motion role.
-3. Implement smallest coherent slice with purposeful motion, reduced-motion behavior, keyboard operation, visible focus, semantic structure, contrast, and labels.
-4. Test narrow/mobile/wide containment. No horizontal leak, clipped text, overlap, or unreadable measure.
+1. Inspect target and nearest pattern.
+2. Lock type, spacing, surfaces, color, asset role, and motion role.
+3. Implement smallest coherent slice with purposeful motion, reduced motion, keyboard operation, visible focus, semantics, contrast, and labels.
+4. Test narrow/mobile/wide containment: no horizontal leak, clipped text, overlap, or unreadable measure.
 5. Check loading, error, and empty states when applicable.
-6. Check interaction and performance proportional to change. Avoid layout shift, oversized media, needless client work.
+6. Check proportional interaction/performance. Avoid layout shift, oversized media, needless client work.
 
 Use real assets when supplied. Never invent research or claim screenshot parity without visual comparison.

package/plugin/skills/programming/SKILL.md

@@ -5,28 +5,37 @@ description: Use when a task edits Python, Rust, TypeScript, Go, or their matchi
 
 # Programming
 
-Before editing `.py`, `.pyi`, `.rs`, `.ts`, `.tsx`, `.mts`, `.cts`, `.go`, or matching manifests, read `references/<language>/README.md`. Load `references/logging.md` only for logs and all `references/rust-ub/` for Rust unsafe or FFI.
+Before editing `.py`, `.pyi`, `.rs`, `.ts`, `.tsx`, `.mts`, `.cts`, `.go`, or matching manifests, read `references/<language>/README.md`. Load `references/logging.md` only for logs; load all `references/rust-ub/` for Rust unsafe/FFI.
 
 ## Core
 
-- Best code: none, existing helper, standard API, native feature, installed dependency, one line, then new code.
+- Prefer, in order: no code, existing helper, standard API, native feature, installed dependency, one line, new code.
 - Trace real flow and callers. Fix shared root seam once.
 - Parse untrusted input once at boundary. Interior stays typed.
-- Make illegal states unrepresentable. Distinct primitives get distinct types.
-- Immutable default. No parameter mutation without explicit purpose.
+- Make illegal states unrepresentable; give distinct primitives distinct types.
+- Default immutable; mutate parameters only with explicit purpose.
 - Exhaustive variant handling.
 - No untyped escape, ignored diagnostics, non-null assertion, unchecked cast, panic/unwrap in library path.
 - Typed errors. Catch only expected errors; unknown errors propagate to boundary.
 - Existing project stack and logger win.
-- No helper for one caller. No speculative abstraction.
+- No one-caller helper or speculative abstraction.
+
+## Reuse
+
+- One behavior, one implementation.
+- Search before writing; reuse or extend the existing implementation.
+- Never copy-paste logic or maintain parallel variants.
+- Put shared behavior in the smallest stable function, method, type, or module at its common ownership seam. Callers pass data, not reimplement policy.
+- Extract repetition when a second caller or copy exists. Keep one-caller mechanics local unless they form a stable domain abstraction.
+- Reusable code needs cohesive names, explicit typed I/O, and public-seam tests. Parameterize real variation; reject caller branches and boolean modes.
 
 ## Test
 
-Defect: add a public-seam regression test first and confirm the intended failure. Explicit test-first request or clearly defined new behavior without adequate proof: test public seam first, confirm failure, implement minimum green, then refactor. Existing tests may lock small covered changes. Do not force red-green for prose, configuration-only work, trivial mechanical edits, or behavior already adequately covered. Use real object or fake before mock. Deterministic fixture; no sleep. Add an end-to-end user outcome for new behavior when proportional. Run smallest test in loop.
+Defect: add a public-seam regression test first and confirm the intended failure. Explicit test-first request or clearly defined new behavior without adequate proof: test public seam, confirm failure, implement minimum green, refactor. Existing tests may lock small covered changes. Do not force red-green for prose, configuration-only work, trivial mechanical edits, or behavior already adequately covered. Prefer real object or fake to mock; use deterministic fixtures, never sleep. Add a proportional end-to-end user outcome for new behavior. Run smallest test in loop.
 
 ## Size
 
-After write, measure pure LOC. Under 200 healthy; 200–250 warn; over 250 split unless generated table or documented indivisible state machine. Function over three independent parameters needs named domain input or specific reason.
+After writing, measure pure LOC: under 200 healthy; 200–250 warn; over 250 split unless generated table or documented indivisible state machine. Over three independent parameters requires named domain input or a specific reason.
 
 ## Finish
 

package/plugin/skills/security-research/SKILL.md

@@ -5,17 +5,17 @@ description: Use when an authorized task needs repository security review, threa
 
 # Security Research
 
-Main agent owns scope, threat surface, dedupe, proof, severity, report. No Team Mode. At most two independent lanes when useful: `explorer` for one internal surface; `librarian` for named standards or dependency facts. No duplicate hunters, recursive delegation, reviewer loop, or full-history fork.
+Main agent owns scope, threat surface, dedupe, proof, severity, report. No Team Mode. When useful, at most two independent lanes: `explorer` for one internal surface; `librarian` for named standards/dependency facts. No duplicate hunters, recursive delegation, reviewer loop, or full-history fork.
 
 ## Rules
 
 - Name target: repo, diff, path, release, or threat surface.
-- Map entry points, attacker input, trust boundaries, sensitive assets, sinks, privilege transitions.
+- Map entry points, attacker input, trust boundaries, assets, sinks, privilege transitions.
 - No severity without reachable attack path.
-- High or critical needs safe local PoC or decisive static proof plus concrete preconditions and impact.
+- High/critical needs safe local PoC or decisive static proof with concrete preconditions and impact.
 - CWE classifies weakness. Severity measures exploitability and impact. Keep separate.
 - Generic hardening is not finding.
-- Never attack real or third-party systems. Use local fixture, toy payload, dry run, or static proof.
+- Never attack real/third-party systems; use local fixture, toy payload, dry run, or static proof.
 
 ## Flow
 
@@ -23,16 +23,16 @@ Main agent owns scope, threat surface, dedupe, proof, severity, report. No Team
 2. Threat surface: attacker capability, controlled input, boundary, sink, asset.
 3. Candidates: title, path/function, attack path, impact, CWE candidate, evidence, safe proof idea.
 4. Deduplicate by root cause and attack path.
-5. Validate strongest candidate. Reproduce, falsify, or downgrade. Record exact command and observed output.
-6. Calibrate severity from proven preconditions, reachability, privilege, user action, scope, confidentiality, integrity, availability.
+5. Validate strongest candidate: reproduce, falsify, or downgrade. Record exact command/output.
+6. Calibrate severity from proven preconditions, reachability, privilege, user action, scope, confidentiality, integrity, and availability.
 7. Give minimum fix and public-seam regression test.
 
 ## Report
 
 Lead with `PASS`, `PASS WITH FINDINGS`, or `BLOCK`.
 
-For each surviving finding: severity, title, CWE, affected path/function, attacker capability, attack path, proof, impact, minimum fix, regression check.
+Each finding: severity, title, CWE, affected path/function, attacker capability, attack path, proof, impact, minimum fix, regression check.
 
-Also list downgraded and rejected candidates with reason. End with residual risk: untested surfaces and why.
+List downgraded/rejected candidates with reasons. End with residual risk: untested surfaces and why.
 
 Use CWE, OWASP WSTG/ASVS, and CVSS v4 only when relevant. Cite exact source. Do not claim precise CVSS score without scoring every metric.