knodin 0.7.6 → 0.8.3

package/dist/src/server.js

@@ -0,0 +1,217 @@
+/**
+ * knodin MCP server (stdio).
+ *
+ * Registration uses the low-level request-handler API — NOT the high-level
+ * `server.tool()` overload — so `tsc` never pays the Zod→handler-arg inference
+ * cost that OOMs large tool sets. Gateway `inputSchema`s are hand-written JSON
+ * Schema literals; handler args are narrowed manually inside each dispatcher.
+ */
+import path from "node:path";
+import { fileURLToPath } from "node:url";
+import { Server } from "@modelcontextprotocol/sdk/server/index.js";
+import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
+import { CallToolRequestSchema, ListToolsRequestSchema } from "@modelcontextprotocol/sdk/types.js";
+import { recordDiagnosticFailure } from "./diagnostics.js";
+import { assertMcpMemoryHeadroom, classifyMcpFailure, createMcpRequestContext, recordMcpLifecycle, superviseMcpRequest, } from "./mcp-reliability.js";
+import { WorkerSupervisor } from "./mcp-worker-supervisor.js";
+import { attachMcpRequestSignal, getKnodinTools, handleKnodinTool } from "./tools/knodin-tools.js";
+import { KNODIN_VERSION } from "./version.js";
+const RETRY_SAFE_OPERATIONS = new Set([
+    "architecture_overview",
+    "context",
+    "docs",
+    "explain",
+    "map",
+    "query",
+    "review",
+    "search",
+    "status",
+]);
+const RECOVERY_LANE_OPERATIONS = new Set(["docs"]);
+class SafeGatewayScheduler {
+    tail = Promise.resolve();
+    async run(operation, signal, task) {
+        if (RECOVERY_LANE_OPERATIONS.has(operation))
+            return task();
+        const predecessor = this.tail;
+        let release;
+        this.tail = new Promise((resolve) => {
+            release = resolve;
+        });
+        try {
+            await predecessor;
+            if (signal.aborted)
+                throw Object.assign(new Error("MCP request cancelled before dispatch"), {
+                    code: "KNODIN_REQUEST_CANCELLED",
+                });
+            return await task();
+        }
+        finally {
+            release();
+        }
+    }
+}
+function ignoreNotificationFailure() {
+    // Progress is advisory; a client that stops accepting it must not fail the tool call.
+}
+function startHeartbeat(notify, operation, startedAt) {
+    return setInterval(() => {
+        void notify(`${operation} still running (${Math.round((Date.now() - startedAt) / 1000)}s)`).catch(ignoreNotificationFailure);
+    }, 10_000);
+}
+export function recordMcpDiagnosticFailure(args, error, defaultRepo = process.cwd()) {
+    const effectiveRepo = typeof args?.repoPath === "string" ? args.repoPath : defaultRepo;
+    const diagnostic = recordDiagnosticFailure(effectiveRepo, {
+        surface: "mcp",
+        operation: typeof args?.operation === "string" ? args.operation : "unknown",
+        phase: "dispatch",
+        error,
+    });
+    const message = error instanceof Error ? error.message : String(error);
+    return diagnostic.recorded
+        ? new Error(`${message} [diagnostic ${diagnostic.correlationId}]`, { cause: error })
+        : error;
+}
+export function createServer(dispatch = handleKnodinTool) {
+    const scheduler = new SafeGatewayScheduler();
+    const workers = new WorkerSupervisor();
+    const server = new Server({ name: "knodin", version: KNODIN_VERSION }, {
+        capabilities: { tools: {} },
+        instructions: "Use knodin first for cold or unfamiliar codebase work: context to orient, explain for source-evidenced symbol context, query impact before meaningful edits, and review before handoff. Prefer direct reads for exact literals, non-code files, or files just edited this turn. If status reports repair-needed, follow its typed repair steps: repair graph damage or init displaced lifecycle routing before relying on graph evidence.",
+    });
+    server.onclose = () => {
+        void workers.close();
+    };
+    server.setRequestHandler(ListToolsRequestSchema, async () => ({
+        tools: getKnodinTools(),
+    }));
+    server.setRequestHandler(CallToolRequestSchema, async (request, extra) => {
+        const { name, arguments: args } = request.params;
+        if (name !== "knodin") {
+            throw new Error(`unknown tool: ${name}`);
+        }
+        const values = args;
+        const operation = typeof values?.operation === "string" ? values.operation : "unknown";
+        const progressToken = extra._meta?.progressToken;
+        const requestId = String(extra.requestId);
+        const startedAt = Date.now();
+        const repoPath = typeof values?.repoPath === "string" ? values.repoPath : process.cwd();
+        const context = createMcpRequestContext(requestId, operation, repoPath);
+        let progressSequence = 0;
+        let heartbeat;
+        recordMcpLifecycle(context, "start");
+        try {
+            assertMcpMemoryHeadroom();
+            const supervised = dispatch === handleKnodinTool && operation !== "docs";
+            const execution = supervised
+                ? await workers.worker(repoPath).execute(context, args, extra.signal, (sequence, phase) => {
+                    progressSequence = sequence;
+                    recordMcpLifecycle(context, "progress", { sequence });
+                    if (progressToken !== undefined)
+                        void extra
+                            .sendNotification({
+                            method: "notifications/progress",
+                            params: { progressToken, progress: sequence, message: phase },
+                        })
+                            .catch(ignoreNotificationFailure);
+                })
+                : {
+                    result: await scheduler.run(operation, extra.signal, () => superviseMcpRequest(context, extra.signal, async (operationSignal) => {
+                        const notify = async (message) => {
+                            if (progressToken === undefined)
+                                return;
+                            progressSequence++;
+                            recordMcpLifecycle(context, "progress", { sequence: progressSequence });
+                            await extra.sendNotification({
+                                method: "notifications/progress",
+                                params: {
+                                    progressToken,
+                                    progress: progressSequence,
+                                    message,
+                                },
+                            });
+                        };
+                        await notify(`Starting ${operation} (deadline ${context.deadlineMs}ms)`);
+                        heartbeat = startHeartbeat(notify, operation, startedAt);
+                        // Keep the production handler explicit: knodin's own source-evidence
+                        // extractor maps this low-level MCP registration to its dispatcher.
+                        if (dispatch === handleKnodinTool) {
+                            attachMcpRequestSignal(args, operationSignal);
+                            return await handleKnodinTool(args);
+                        }
+                        return await dispatch(args, { signal: operationSignal });
+                    }, { cooperativeCancellation: operation === "repair" })),
+                };
+            const result = execution.result;
+            recordMcpLifecycle(context, "success", { elapsedMs: Date.now() - startedAt });
+            return {
+                _meta: {
+                    requestId: context.requestId,
+                    traceId: context.traceId,
+                    ...("predecessorTraceId" in execution
+                        ? { predecessorTraceId: execution.predecessorTraceId }
+                        : {}),
+                },
+                content: [{ type: "text", text: JSON.stringify(result, null, 2) }],
+            };
+        }
+        catch (error) {
+            const failure = classifyMcpFailure(error);
+            recordMcpLifecycle(context, "failure", {
+                kind: failure.kind,
+                elapsedMs: Date.now() - startedAt,
+            });
+            const recorded = recordMcpDiagnosticFailure(values, error);
+            const message = recorded instanceof Error ? recorded.message : String(recorded);
+            const retrySafe = RETRY_SAFE_OPERATIONS.has(operation);
+            return {
+                isError: true,
+                _meta: { requestId: context.requestId, traceId: context.traceId },
+                content: [
+                    {
+                        type: "text",
+                        text: JSON.stringify({
+                            code: failure.code,
+                            kind: failure.kind,
+                            message,
+                            traceId: context.traceId,
+                            requestId,
+                            operation,
+                            phase: "dispatch",
+                            deadlineMs: context.deadlineMs,
+                            retrySafe,
+                            diagnosticLog: ".knodin/diagnostics/events.jsonl (only when opt-in diagnostics are enabled)",
+                            recovery: retrySafe
+                                ? "Retry once. If the transport closes, restart the MCP client and run `knodin doctor`; include `knodin diagnostics export` when diagnostics are enabled."
+                                : "Do not retry automatically. Run `knodin doctor`, inspect repository state, and include `knodin diagnostics export` when diagnostics are enabled.",
+                        }, null, 2),
+                    },
+                ],
+            };
+        }
+        finally {
+            if (heartbeat)
+                clearInterval(heartbeat);
+            recordMcpLifecycle(context, "cleanup", { elapsedMs: Date.now() - startedAt });
+        }
+    });
+    return server;
+}
+async function main() {
+    const server = createServer();
+    const transport = new StdioServerTransport();
+    await server.connect(transport);
+    // stdio server: log to stderr so stdout stays a clean JSON-RPC channel.
+    console.error("knodin MCP server running on stdio");
+}
+/** Start the stdio gateway from either the direct server entry or the packaged CLI. */
+export async function startServer() {
+    await main();
+}
+// Run only when invoked directly (not when imported by tests).
+if (process.argv[1] && fileURLToPath(import.meta.url) === path.resolve(process.argv[1])) {
+    startServer().catch((err) => {
+        console.error(err);
+        process.exit(1);
+    });
+}

package/dist/src/structural-fast-path.js

@@ -0,0 +1,344 @@
+import fs from "node:fs";
+import path from "node:path";
+import { compareBytes } from "./compare.js";
+import { contentFingerprint, readStructuralSnapshot, } from "./structural-snapshot.js";
+function exactFile(repo, file) {
+    try {
+        return (contentFingerprint(fs.readFileSync(path.resolve(repo, file.path))) === file.contentFingerprint);
+    }
+    catch {
+        return false;
+    }
+}
+const DIRECT_FILE_EXTENSIONS = new Set([
+    ".cjs",
+    ".cs",
+    ".cts",
+    ".go",
+    ".java",
+    ".js",
+    ".jsx",
+    ".mjs",
+    ".mts",
+    ".py",
+    ".rs",
+    ".ts",
+    ".tsx",
+]);
+function lexicalEndLine(lines, startIndex, python) {
+    if (python) {
+        const indent = /^\s*/.exec(lines[startIndex])?.[0].length ?? 0;
+        let end = startIndex + 1;
+        for (let index = startIndex + 1; index < lines.length; index++) {
+            const line = lines[index];
+            if (line.trim() && (/^\s*/.exec(line)?.[0].length ?? 0) <= indent)
+                break;
+            end = index + 1;
+        }
+        return end;
+    }
+    let depth = 0;
+    let opened = false;
+    for (let index = startIndex; index < lines.length; index++) {
+        for (const character of lines[index]) {
+            if (character === "{") {
+                opened = true;
+                depth++;
+            }
+            else if (character === "}")
+                depth--;
+        }
+        if (opened && depth <= 0)
+            return index + 1;
+        if (!opened && /[;}]\s*$/.test(lines[index]))
+            return index + 1;
+    }
+    return startIndex + 1;
+}
+function genericFunctionSymbol(line) {
+    const opening = line.indexOf("(");
+    if (opening <= 0)
+        return null;
+    let depth = 0;
+    let closing = -1;
+    for (let index = opening; index < line.length; index += 1) {
+        if (line[index] === "(")
+            depth += 1;
+        if (line[index] !== ")")
+            continue;
+        depth -= 1;
+        if (depth === 0) {
+            closing = index;
+            break;
+        }
+    }
+    if (closing < 0)
+        return null;
+    const suffix = line.slice(closing + 1).trimStart();
+    if (!suffix.startsWith("{") && !suffix.startsWith("=>"))
+        return null;
+    const prefix = line.slice(0, opening).trim();
+    if (/[=;{}()]/.test(prefix))
+        return null;
+    const tokens = prefix.split(/\s+/);
+    const symbol = tokens.at(-1) ?? "";
+    return tokens.length >= 2 && /^[A-Za-z_$][\w$]*$/.test(symbol) ? symbol : null;
+}
+function firstPatternSymbol(line, patterns) {
+    for (const { kind, expression } of patterns) {
+        const match = expression.exec(line);
+        if (match)
+            return { kind, symbol: match[1] };
+    }
+    return null;
+}
+function directSymbols(filePath, content) {
+    const extension = path.extname(filePath).toLowerCase();
+    const python = extension === ".py";
+    const lines = content.split(/\r?\n/);
+    const symbols = [];
+    const patterns = python
+        ? [
+            { kind: "class", expression: /^\s*class\s+([A-Za-z_$][\w$]*)\b/ },
+            { kind: "function", expression: /^\s*(?:async\s+)?def\s+([A-Za-z_$][\w$]*)\s*\(/ },
+        ]
+        : [
+            {
+                kind: "class",
+                expression: /^\s*(?:(?:export|public|private|protected|internal|abstract|final|sealed|static)\s+)*(?:class|interface|enum|struct|trait|record|type)\s+([A-Za-z_$][\w$]*)\b/,
+            },
+            {
+                kind: "function",
+                expression: /^\s*(?:(?:export|public|private|protected|internal|async|static|final|pub(?:\([^)]*\))?)\s+)*(?:func|fn|function)\s+([A-Za-z_$][\w$]*)\s*\(/,
+            },
+            {
+                kind: "function",
+                expression: /^\s*(?:(?:export|const|let|var|public|private|protected|static|readonly)\s+)*([A-Za-z_$][\w$]*)\s*=\s*(?:async\s*)?(?:\([^)]*\)|[A-Za-z_$][\w$]*)\s*=>/,
+            },
+        ];
+    for (const [index, line] of lines.entries()) {
+        let found = firstPatternSymbol(line, patterns);
+        if (!found && !python) {
+            const symbol = genericFunctionSymbol(line);
+            if (symbol)
+                found = { kind: "function", symbol };
+        }
+        if (found) {
+            const { kind, symbol } = found;
+            const visibility = /\b(public|private|protected|internal)\b/.exec(line)?.[1] ??
+                (python && symbol.startsWith("_") ? "private" : "default");
+            symbols.push({
+                identity: null,
+                symbol,
+                kind,
+                line: index + 1,
+                endLine: lexicalEndLine(lines, index, python),
+                signature: line
+                    .trim()
+                    .split(/\s*(?:\{|=>)\s*/, 1)[0]
+                    .slice(0, 500),
+                visibility,
+                exported: /\bexport\b/.test(line) || /\bpub\b/.test(line),
+                evidenceQuality: "lexical-structural-fallback",
+            });
+        }
+    }
+    return symbols;
+}
+function directFile(repo, filePath) {
+    if (!DIRECT_FILE_EXTENSIONS.has(path.extname(filePath).toLowerCase()))
+        return null;
+    const root = fs.realpathSync(repo);
+    const absolute = path.resolve(root, filePath);
+    let real;
+    try {
+        real = fs.realpathSync(absolute);
+    }
+    catch {
+        return null;
+    }
+    const relative = path.relative(root, real);
+    if (relative === ".." || relative.startsWith(`..${path.sep}`) || path.isAbsolute(relative))
+        return null;
+    const content = fs.readFileSync(real, "utf8");
+    return {
+        path: relative.split(path.sep).join("/"),
+        sizeBytes: Buffer.byteLength(content),
+        mtimeMs: fs.statSync(real).mtimeMs,
+        contentFingerprint: contentFingerprint(content),
+        symbols: directSymbols(filePath, content),
+    };
+}
+export function runStructuralFastPath(repo, pattern, target, limit = 100) {
+    const snapshot = readStructuralSnapshot(repo);
+    if (!snapshot && pattern !== "file_summary")
+        return null;
+    let matching = (snapshot?.files ?? []).filter((file) => !target || file.path.includes(target));
+    if (pattern === "file_summary") {
+        const indexed = matching.find((file) => file.path === target);
+        if (!indexed || !exactFile(repo, indexed)) {
+            const direct = directFile(repo, target);
+            if (!direct)
+                return null;
+            matching = [direct];
+        }
+    }
+    else if (pattern === "batch_outline") {
+        matching = matching.flatMap((file) => {
+            if (exactFile(repo, file))
+                return [file];
+            const direct = directFile(repo, file.path);
+            return direct ? [direct] : [];
+        });
+    }
+    let rows = [];
+    let fingerprint;
+    let fresh = false;
+    if (pattern === "project_overview") {
+        fresh = matching.every((file) => {
+            try {
+                const stat = fs.statSync(path.resolve(repo, file.path));
+                return stat.size === file.sizeBytes && stat.mtimeMs === file.mtimeMs;
+            }
+            catch {
+                return false;
+            }
+        });
+        const directories = new Map();
+        for (const file of matching) {
+            const directory = file.path.includes("/") ? file.path.split("/", 1)[0] : ".";
+            const value = directories.get(directory) ?? { fileCount: 0, sizeBytes: 0 };
+            value.fileCount += 1;
+            value.sizeBytes += file.sizeBytes;
+            directories.set(directory, value);
+        }
+        rows = [...directories]
+            // Byte order: this sort decides which directories survive the
+            // `rows.slice(0, limit)` below, not merely how they read.
+            .sort(([a], [b]) => compareBytes(a, b))
+            .map(([symbol, value]) => ({ symbol, kind: "directory", ...value }));
+        fingerprint = contentFingerprint(JSON.stringify(matching.map(({ path: file, contentFingerprint: hash }) => [file, hash])));
+    }
+    else {
+        const files = pattern === "file_summary" ? matching.filter((file) => file.path === target) : matching;
+        fresh = files.length > 0 && files.every((file) => exactFile(repo, file));
+        fingerprint =
+            files.length === 1
+                ? files[0].contentFingerprint
+                : contentFingerprint(JSON.stringify(files.map((file) => [file.path, file.contentFingerprint])));
+        rows = files.flatMap((file) => file.symbols.map((symbol) => ({ ...symbol, file: file.path })));
+    }
+    const selected = rows.slice(0, limit);
+    let upgrade;
+    if (pattern === "file_summary")
+        upgrade = { operation: "explain", target };
+    else if (pattern === "project_overview")
+        upgrade = { operation: "query", pattern: "architecture_overview", target: "" };
+    else
+        upgrade = { operation: "query", pattern: "impact", target };
+    return {
+        pattern,
+        target,
+        count: selected.length,
+        results: selected,
+        ...(rows.length > selected.length ? { hasMore: true } : {}),
+        staleness: fresh ? "fresh" : "unknown",
+        graphEnriched: false,
+        evidence: {
+            ...(pattern === "file_summary" ? { path: target } : {}),
+            contentFingerprint: fingerprint,
+            evidenceQuality: matching.some((file) => file.symbols.some((symbol) => symbol.evidenceQuality === "lexical-structural-fallback"))
+                ? "direct-lexical-parser"
+                : "indexed-structural-snapshot",
+            uniqueness: pattern === "project_overview" ? "not-applicable" : "file-scoped-only",
+            evidenceGeneratedAt: snapshot?.generatedAt ?? new Date().toISOString(),
+        },
+        upgrade,
+    };
+}
+export function runExactSymbolFastPath(repo, symbol, filePath) {
+    const snapshot = readStructuralSnapshot(repo);
+    const indexed = snapshot?.files.find((candidate) => candidate.path === filePath);
+    const file = indexed && exactFile(repo, indexed) ? indexed : directFile(repo, filePath);
+    if (!file)
+        return null;
+    const matches = file.symbols.filter((candidate) => candidate.symbol === symbol);
+    if (matches.length === 0)
+        return null;
+    const fresh = exactFile(repo, file);
+    const evidence = {
+        path: filePath,
+        contentFingerprint: file.contentFingerprint,
+        evidenceQuality: file.symbols.some((candidate) => candidate.evidenceQuality === "lexical-structural-fallback")
+            ? "direct-lexical-parser"
+            : "indexed-structural-snapshot",
+        uniqueness: "file-scoped-only",
+        evidenceGeneratedAt: snapshot?.generatedAt ?? new Date().toISOString(),
+    };
+    if (matches.length > 1) {
+        return {
+            symbol,
+            file: filePath,
+            ambiguity: {
+                candidates: matches.map((match) => ({
+                    identity: match.identity ?? "",
+                    file: filePath,
+                    kind: match.kind,
+                    line: match.line,
+                })),
+            },
+            staleness: fresh ? "fresh" : "unknown",
+            graphEnriched: false,
+            evidence,
+            upgrade: { operation: "explain", target: symbol },
+        };
+    }
+    const match = matches[0];
+    const lines = fs.readFileSync(path.resolve(repo, filePath), "utf8").split(/\r?\n/);
+    const source = fresh
+        ? lines
+            .slice(match.line - 1, match.endLine)
+            .map((line, offset) => `${match.line + offset}: ${line}`)
+            .join("\n")
+        : undefined;
+    return {
+        symbol,
+        ...(match.identity ? { identity: match.identity } : {}),
+        file: filePath,
+        ...(source ? { source } : {}),
+        staleness: fresh ? "fresh" : "unknown",
+        graphEnriched: false,
+        evidence,
+        upgrade: { operation: "explain", target: symbol },
+    };
+}
+export function tryStructuralFastPath(argv, cwd) {
+    const valueAfter = (flag) => {
+        const index = argv.indexOf(flag);
+        return index >= 0
+            ? argv[index + 1]
+            : argv.find((value) => value.startsWith(`${flag}=`))?.slice(flag.length + 1);
+    };
+    if (argv[0] === "explain" && argv[1] && valueAfter("--file")) {
+        const repo = path.resolve(valueAfter("--repo") ?? cwd ?? process.cwd());
+        const result = runExactSymbolFastPath(repo, argv[1], valueAfter("--file"));
+        if (!result)
+            return false;
+        process.stdout.write(`${JSON.stringify(result)}\n`);
+        return true;
+    }
+    if (argv[0] !== "query" ||
+        !["file_summary", "batch_outline", "project_overview"].includes(argv[1] ?? ""))
+        return false;
+    const pattern = argv[1];
+    const repo = path.resolve(valueAfter("--repo") ?? cwd ?? process.cwd());
+    const target = pattern === "project_overview" ? "" : (argv[2] ?? "");
+    const limit = Number(valueAfter("--limit") ?? 100);
+    if (!target && pattern !== "project_overview")
+        return false;
+    const result = runStructuralFastPath(repo, pattern, target, Number.isInteger(limit) && limit > 0 ? limit : 100);
+    if (!result)
+        return false;
+    process.stdout.write(`${JSON.stringify(result)}\n`);
+    return true;
+}

package/dist/src/structural-snapshot.js

@@ -0,0 +1,37 @@
+import crypto from "node:crypto";
+import fs from "node:fs";
+import path from "node:path";
+import { compareBytes } from "./compare.js";
+import { resolveStateDir } from "./engine/state-paths.js";
+export const STRUCTURAL_SNAPSHOT_VERSION = 1;
+export function contentFingerprint(content) {
+    return `sha256:${crypto.createHash("sha256").update(content).digest("hex")}`;
+}
+export function structuralSnapshotPath(repo) {
+    return path.join(resolveStateDir(repo), "structural-v1.json");
+}
+export function writeStructuralSnapshot(repo, files) {
+    const destination = structuralSnapshotPath(repo);
+    fs.mkdirSync(path.dirname(destination), { recursive: true });
+    const snapshot = {
+        schemaVersion: STRUCTURAL_SNAPSHOT_VERSION,
+        generatedAt: new Date().toISOString(),
+        // Byte order, not collation: this array is written to disk and read back
+        // on any host, so the snapshot's bytes must not depend on the locale.
+        files: [...files].sort((left, right) => compareBytes(left.path, right.path)),
+    };
+    const temporary = `${destination}.${process.pid}.tmp`;
+    fs.writeFileSync(temporary, `${JSON.stringify(snapshot)}\n`, { mode: 0o600 });
+    fs.renameSync(temporary, destination);
+}
+export function readStructuralSnapshot(repo) {
+    try {
+        const parsed = JSON.parse(fs.readFileSync(structuralSnapshotPath(repo), "utf8"));
+        return parsed?.schemaVersion === STRUCTURAL_SNAPSHOT_VERSION && Array.isArray(parsed.files)
+            ? parsed
+            : null;
+    }
+    catch {
+        return null;
+    }
+}