@sweny-ai/core 0.1.76 → 0.1.78

package/dist/__tests__/claude.test.js

@@ -1,427 +0,0 @@
-import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
-// ─── tryParseJSON tests (via inlined copy since it's private) ────
-describe("JSON extraction strategies", () => {
-    function tryParseJSON(text) {
-        if (!text)
-            return {};
-        const codeBlockMatch = text.match(/```(?:json)?\s*([\s\S]*?)```/);
-        if (codeBlockMatch) {
-            try {
-                const parsed = JSON.parse(codeBlockMatch[1].trim());
-                if (typeof parsed === "object" && parsed !== null)
-                    return parsed;
-            }
-            catch {
-                /* try next */
-            }
-        }
-        const lastBrace = text.lastIndexOf("}");
-        if (lastBrace !== -1) {
-            let depth = 0;
-            for (let i = lastBrace; i >= 0; i--) {
-                if (text[i] === "}")
-                    depth++;
-                else if (text[i] === "{")
-                    depth--;
-                if (depth === 0) {
-                    try {
-                        const parsed = JSON.parse(text.slice(i, lastBrace + 1));
-                        if (typeof parsed === "object" && parsed !== null)
-                            return parsed;
-                    }
-                    catch {
-                        /* try next */
-                    }
-                    break;
-                }
-            }
-        }
-        try {
-            const parsed = JSON.parse(text.trim());
-            if (typeof parsed === "object" && parsed !== null)
-                return parsed;
-        }
-        catch {
-            /* fall through */
-        }
-        return {};
-    }
-    it("extracts JSON from code block", () => {
-        const text = 'Here are the results:\n```json\n{"severity": "high", "count": 42}\n```';
-        expect(tryParseJSON(text)).toEqual({ severity: "high", count: 42 });
-    });
-    it("extracts JSON from code block without json label", () => {
-        const text = 'Results:\n```\n{"ok": true}\n```';
-        expect(tryParseJSON(text)).toEqual({ ok: true });
-    });
-    it("extracts JSON from inline brace block", () => {
-        const text = 'The analysis is complete. {"result": "success", "items": [1,2,3]}';
-        expect(tryParseJSON(text)).toEqual({ result: "success", items: [1, 2, 3] });
-    });
-    it("extracts last JSON object when multiple exist", () => {
-        const text = 'First: {"a": 1}\nThen: {"b": 2}';
-        expect(tryParseJSON(text)).toEqual({ b: 2 });
-    });
-    it("handles nested JSON objects", () => {
-        const text = 'Output: {"outer": {"inner": true}}';
-        expect(tryParseJSON(text)).toEqual({ outer: { inner: true } });
-    });
-    it("parses pure JSON text", () => {
-        const text = '{"pure": "json"}';
-        expect(tryParseJSON(text)).toEqual({ pure: "json" });
-    });
-    it("returns empty object for no JSON", () => {
-        expect(tryParseJSON("No JSON here at all")).toEqual({});
-    });
-    it("returns empty object for invalid JSON", () => {
-        expect(tryParseJSON("{invalid json}")).toEqual({});
-    });
-    it("handles JSON with braces in strings", () => {
-        const text = '{"message": "use {curly} braces"}';
-        expect(tryParseJSON(text)).toEqual({ message: "use {curly} braces" });
-    });
-    it("handles multiline JSON in code block", () => {
-        const text = '```json\n{\n  "a": 1,\n  "b": {\n    "c": 2\n  }\n}\n```';
-        expect(tryParseJSON(text)).toEqual({ a: 1, b: { c: 2 } });
-    });
-    it("prefers code block over inline JSON", () => {
-        const text = '{"inline": true}\n```json\n{"codeblock": true}\n```';
-        expect(tryParseJSON(text)).toEqual({ codeblock: true });
-    });
-    it("handles empty code block gracefully", () => {
-        const text = '```json\n\n```\n{"fallback": true}';
-        expect(tryParseJSON(text)).toEqual({ fallback: true });
-    });
-});
-// ─── ClaudeClient tests ──────────────────────────────────────────
-// Mock the agent SDK to test ClaudeClient behavior
-describe("ClaudeClient", () => {
-    let mockQuery;
-    let mockCreateSdkMcpServer;
-    let mockSdkTool;
-    let ClaudeClient;
-    /** Helper: create an async generator that yields the given messages */
-    function makeStream(messages) {
-        return (async function* () {
-            for (const msg of messages) {
-                yield msg;
-            }
-        })();
-    }
-    beforeEach(async () => {
-        mockQuery = vi.fn();
-        mockCreateSdkMcpServer = vi.fn().mockReturnValue({ type: "sdk", name: "sweny-core" });
-        mockSdkTool = vi.fn().mockImplementation((name, desc, schema, handler) => ({
-            name,
-            description: desc,
-            inputSchema: schema,
-            handler,
-        }));
-        vi.doMock("@anthropic-ai/claude-agent-sdk", () => ({
-            query: mockQuery,
-            createSdkMcpServer: mockCreateSdkMcpServer,
-            tool: mockSdkTool,
-        }));
-        const mod = await import("../claude.js");
-        ClaudeClient = mod.ClaudeClient;
-    });
-    afterEach(() => {
-        vi.restoreAllMocks();
-        vi.resetModules();
-    });
-    it("sends instruction and context in prompt", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: '{"done": true}' }]));
-        const client = new ClaudeClient();
-        await client.run({
-            instruction: "Analyze the logs",
-            context: { alert: "cpu spike" },
-            tools: [],
-        });
-        const callArgs = mockQuery.mock.calls[0][0];
-        expect(callArgs.prompt).toContain("Analyze the logs");
-        expect(callArgs.prompt).toContain("cpu spike");
-    });
-    it("passes system prompt and permissionMode", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "ok" }]));
-        const client = new ClaudeClient();
-        await client.run({ instruction: "x", context: {}, tools: [] });
-        const opts = mockQuery.mock.calls[0][0].options;
-        expect(opts.systemPrompt).toContain("automated workflow");
-        expect(opts.permissionMode).toBe("bypassPermissions");
-    });
-    it("returns success with parsed JSON data", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([
-            { type: "result", subtype: "success", result: 'Analysis complete.\n```json\n{"severity": "high"}\n```' },
-        ]));
-        const client = new ClaudeClient();
-        const result = await client.run({
-            instruction: "Check severity",
-            context: {},
-            tools: [],
-        });
-        expect(result.status).toBe("success");
-        expect(result.data.severity).toBe("high");
-    });
-    it("creates MCP server with converted tools", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "done" }]));
-        const handler = vi.fn().mockResolvedValue({ ok: true });
-        const client = new ClaudeClient();
-        await client.run({
-            instruction: "Do work",
-            context: {},
-            tools: [
-                {
-                    name: "my_tool",
-                    description: "A test tool",
-                    input_schema: {
-                        type: "object",
-                        properties: { query: { type: "string", description: "Search query" } },
-                        required: ["query"],
-                    },
-                    handler,
-                },
-            ],
-        });
-        // tool() was called for the tool
-        expect(mockSdkTool).toHaveBeenCalledOnce();
-        expect(mockSdkTool.mock.calls[0][0]).toBe("my_tool");
-        expect(mockSdkTool.mock.calls[0][1]).toBe("A test tool");
-        // MCP server was created
-        expect(mockCreateSdkMcpServer).toHaveBeenCalledOnce();
-        // mcpServers was passed to query
-        const opts = mockQuery.mock.calls[0][0].options;
-        expect(opts.mcpServers).toBeDefined();
-        expect(opts.mcpServers["sweny-core"]).toBeDefined();
-    });
-    it("does not create MCP server when no tools", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "done" }]));
-        const client = new ClaudeClient();
-        await client.run({ instruction: "x", context: {}, tools: [] });
-        const opts = mockQuery.mock.calls[0][0].options;
-        expect(opts.mcpServers).toBeUndefined();
-    });
-    it("tracks tool calls via handler wrapper", async () => {
-        // Simulate: query returns success, but the tool handler was invoked by MCP internally
-        // We test the handler wrapper directly since query is mocked
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "done" }]));
-        const handler = vi.fn().mockResolvedValue({ items: [1, 2] });
-        const client = new ClaudeClient();
-        const result = await client.run({
-            instruction: "x",
-            context: {},
-            tools: [{ name: "fetch_data", description: "d", input_schema: {}, handler }],
-        });
-        // The SDK tool handler was created — invoke it directly to test wrapping
-        const sdkHandler = mockSdkTool.mock.calls[0][3];
-        const callResult = await sdkHandler({ q: "test" }, {});
-        expect(callResult.content[0].text).toContain("items");
-        expect(callResult.isError).toBeUndefined();
-        // Tool call should be tracked
-        expect(result.toolCalls.length).toBeGreaterThanOrEqual(0); // stream didn't trigger it, but direct call did
-    });
-    it("wraps tool handler errors as isError", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "done" }]));
-        const handler = vi.fn().mockRejectedValue(new Error("tool broke"));
-        const client = new ClaudeClient();
-        await client.run({
-            instruction: "x",
-            context: {},
-            tools: [{ name: "bad_tool", description: "d", input_schema: {}, handler }],
-        });
-        // Invoke the SDK handler directly
-        const sdkHandler = mockSdkTool.mock.calls[0][3];
-        const callResult = await sdkHandler({}, {});
-        expect(callResult.content[0].text).toContain("tool broke");
-        expect(callResult.isError).toBe(true);
-    });
-    it("returns failed on query error", async () => {
-        mockQuery.mockReturnValueOnce((async function* () {
-            throw new Error("connection refused");
-        })());
-        const client = new ClaudeClient();
-        const result = await client.run({ instruction: "x", context: {}, tools: [] });
-        expect(result.status).toBe("failed");
-        expect(result.data.error).toContain("connection refused");
-    });
-    it("returns failed on error result subtype", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "error_max_turns", errors: ["Too many turns"], is_error: true }]));
-        const client = new ClaudeClient();
-        const result = await client.run({ instruction: "x", context: {}, tools: [] });
-        expect(result.status).toBe("failed");
-        expect(result.data.error).toContain("Too many turns");
-    });
-    // terminal_reason was added in @anthropic-ai/claude-agent-sdk v0.2.91.
-    // When the turn budget is exhausted the SDK now emits subtype='success'
-    // alongside terminal_reason='max_turns' instead of the old error subtype,
-    // so the claude.ts handler has to look at both fields or silently accept
-    // truncated output as a success.
-    describe("terminal_reason handling (claude-agent-sdk v0.2.91+)", () => {
-        it("returns failed when terminal_reason is max_turns", async () => {
-            mockQuery.mockReturnValueOnce(makeStream([
-                {
-                    type: "result",
-                    subtype: "success",
-                    result: "partial output",
-                    terminal_reason: "max_turns",
-                },
-            ]));
-            const client = new ClaudeClient();
-            const result = await client.run({ instruction: "x", context: {}, tools: [] });
-            expect(result.status).toBe("failed");
-            expect(result.data.error).toContain("max_turns");
-            expect(result.data.error).toContain("terminated early");
-        });
-        it("returns failed for any non-completed terminal_reason", async () => {
-            // Future SDK versions may add new terminal reasons (aborted_tools,
-            // blocking_limit, etc.) — treat anything that isn't 'completed' as a
-            // failure rather than assume it's recoverable.
-            mockQuery.mockReturnValueOnce(makeStream([
-                {
-                    type: "result",
-                    subtype: "success",
-                    result: "partial",
-                    terminal_reason: "aborted_tools",
-                },
-            ]));
-            const client = new ClaudeClient();
-            const result = await client.run({ instruction: "x", context: {}, tools: [] });
-            expect(result.status).toBe("failed");
-            expect(result.data.error).toContain("aborted_tools");
-        });
-        it("returns success when terminal_reason is 'completed'", async () => {
-            mockQuery.mockReturnValueOnce(makeStream([
-                {
-                    type: "result",
-                    subtype: "success",
-                    result: '{"ok": true}',
-                    terminal_reason: "completed",
-                },
-            ]));
-            const client = new ClaudeClient();
-            const result = await client.run({ instruction: "x", context: {}, tools: [] });
-            expect(result.status).toBe("success");
-            expect(result.data.ok).toBe(true);
-        });
-        it("returns success when terminal_reason is absent (pre-v0.2.91 SDK)", async () => {
-            // Old SDKs never set terminal_reason. The guard must not treat an
-            // absent field as a failure or every caller would break on downgrade.
-            mockQuery.mockReturnValueOnce(makeStream([
-                {
-                    type: "result",
-                    subtype: "success",
-                    result: '{"done": true}',
-                    // terminal_reason deliberately omitted
-                },
-            ]));
-            const client = new ClaudeClient();
-            const result = await client.run({ instruction: "x", context: {}, tools: [] });
-            expect(result.status).toBe("success");
-            expect(result.data.done).toBe(true);
-        });
-        it("preserves toolCalls when failing on terminal_reason", async () => {
-            // A max_turns failure should still surface any tool calls that
-            // happened before the budget ran out, so debuggers can see what
-            // Claude was trying to do.
-            mockQuery.mockReturnValueOnce(makeStream([
-                {
-                    type: "assistant",
-                    message: {
-                        content: [
-                            {
-                                type: "tool_use",
-                                name: "search_code",
-                                input: { query: "foo" },
-                            },
-                        ],
-                    },
-                },
-                {
-                    type: "result",
-                    subtype: "success",
-                    result: "partial",
-                    terminal_reason: "max_turns",
-                },
-            ]));
-            const client = new ClaudeClient();
-            const result = await client.run({ instruction: "x", context: {}, tools: [] });
-            expect(result.status).toBe("failed");
-            expect(result.toolCalls).toHaveLength(1);
-            expect(result.toolCalls[0].tool).toBe("search_code");
-        });
-    });
-    it("passes model option to query", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "ok" }]));
-        const client = new ClaudeClient({ model: "claude-sonnet-4-20250514" });
-        await client.run({ instruction: "x", context: {}, tools: [] });
-        expect(mockQuery.mock.calls[0][0].options.model).toBe("claude-sonnet-4-20250514");
-    });
-    it("evaluate returns exact match", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "handle_high" }]));
-        const client = new ClaudeClient();
-        const result = await client.evaluate({
-            question: "Which path?",
-            context: {},
-            choices: [
-                { id: "handle_high", description: "High severity" },
-                { id: "handle_low", description: "Low severity" },
-            ],
-        });
-        expect(result).toBe("handle_high");
-    });
-    it("evaluate returns fuzzy match", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([
-            { type: "result", subtype: "success", result: "I think the answer is handle_low based on the data" },
-        ]));
-        const client = new ClaudeClient();
-        const result = await client.evaluate({
-            question: "Which path?",
-            context: {},
-            choices: [
-                { id: "handle_high", description: "High" },
-                { id: "handle_low", description: "Low" },
-            ],
-        });
-        expect(result).toBe("handle_low");
-    });
-    it("evaluate falls back to first choice", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: "I'm not sure" }]));
-        const client = new ClaudeClient();
-        const result = await client.evaluate({
-            question: "Which?",
-            context: {},
-            choices: [
-                { id: "a", description: "A" },
-                { id: "b", description: "B" },
-            ],
-        });
-        expect(result).toBe("a");
-    });
-    it("evaluate falls back on query error", async () => {
-        mockQuery.mockReturnValueOnce((async function* () {
-            throw new Error("offline");
-        })());
-        const client = new ClaudeClient();
-        const result = await client.evaluate({
-            question: "Which?",
-            context: {},
-            choices: [
-                { id: "fallback", description: "Fallback" },
-                { id: "other", description: "Other" },
-            ],
-        });
-        expect(result).toBe("fallback");
-    });
-    it("includes output schema in prompt when provided", async () => {
-        mockQuery.mockReturnValueOnce(makeStream([{ type: "result", subtype: "success", result: '{"severity": "high"}' }]));
-        const client = new ClaudeClient();
-        await client.run({
-            instruction: "Analyze",
-            context: {},
-            tools: [],
-            outputSchema: { type: "object", properties: { severity: { type: "string" } } },
-        });
-        const prompt = mockQuery.mock.calls[0][0].prompt;
-        expect(prompt).toContain("Required Output");
-        expect(prompt).toContain("severity");
-    });
-});

package/dist/__tests__/custom-loader.test.d.ts

@@ -1 +0,0 @@
-export {};

package/dist/__tests__/custom-loader.test.js

@@ -1,116 +0,0 @@
-import { describe, it, expect, vi, beforeEach } from "vitest";
-import { readdirSync, readFileSync, existsSync } from "node:fs";
-vi.mock("node:fs", () => ({
-    readdirSync: vi.fn(),
-    readFileSync: vi.fn(),
-    existsSync: vi.fn(),
-}));
-const mockedExistsSync = vi.mocked(existsSync);
-const mockedReaddirSync = vi.mocked(readdirSync);
-const mockedReadFileSync = vi.mocked(readFileSync);
-// Import after mocking
-const { discoverSkills } = await import("../skills/custom-loader.js");
-describe("custom-loader", () => {
-    beforeEach(() => {
-        vi.clearAllMocks();
-    });
-    describe("config parsing from SKILL.md frontmatter", () => {
-        it("parses config fields from frontmatter", () => {
-            const skillMd = `---
-name: tax-tool
-description: Processes tax filings
-config:
-  MY_TAX_API_KEY:
-    description: API key for the tax service
-    required: true
-  MY_TAX_REGION:
-    description: Tax region code
-    required: false
----
-Do tax stuff.
-`;
-            mockedExistsSync.mockImplementation((p) => {
-                const s = String(p);
-                return s.endsWith(".sweny/skills") || s.endsWith("tax-tool/SKILL.md");
-            });
-            mockedReaddirSync.mockReturnValue([{ name: "tax-tool", isDirectory: () => true }]);
-            mockedReadFileSync.mockReturnValue(skillMd);
-            const skills = discoverSkills("/fake");
-            expect(skills).toHaveLength(1);
-            const skill = skills[0];
-            expect(skill.id).toBe("tax-tool");
-            expect(skill.config).toEqual({
-                MY_TAX_API_KEY: {
-                    description: "API key for the tax service",
-                    required: true,
-                    env: "MY_TAX_API_KEY",
-                },
-                MY_TAX_REGION: {
-                    description: "Tax region code",
-                    required: false,
-                    env: "MY_TAX_REGION",
-                },
-            });
-        });
-        it("produces config: {} when no config block in frontmatter", () => {
-            const skillMd = `---
-name: simple-skill
-description: A simple skill
----
-Instructions here.
-`;
-            mockedExistsSync.mockImplementation((p) => {
-                const s = String(p);
-                return s.endsWith(".sweny/skills") || s.endsWith("simple-skill/SKILL.md");
-            });
-            mockedReaddirSync.mockReturnValue([{ name: "simple-skill", isDirectory: () => true }]);
-            mockedReadFileSync.mockReturnValue(skillMd);
-            const skills = discoverSkills("/fake");
-            expect(skills).toHaveLength(1);
-            expect(skills[0].config).toEqual({});
-        });
-        it("sets env field to the key name automatically", () => {
-            const skillMd = `---
-name: my-api
-description: API skill
-config:
-  CUSTOM_API_TOKEN:
-    description: Token for the API
-    required: true
----
-Use the API.
-`;
-            mockedExistsSync.mockImplementation((p) => {
-                const s = String(p);
-                return s.endsWith(".sweny/skills") || s.endsWith("my-api/SKILL.md");
-            });
-            mockedReaddirSync.mockReturnValue([{ name: "my-api", isDirectory: () => true }]);
-            mockedReadFileSync.mockReturnValue(skillMd);
-            const skills = discoverSkills("/fake");
-            expect(skills[0].config.CUSTOM_API_TOKEN.env).toBe("CUSTOM_API_TOKEN");
-        });
-        it("handles config entries with missing description gracefully", () => {
-            const skillMd = `---
-name: sparse-config
-description: Sparse config skill
-config:
-  SOME_KEY:
-    required: true
----
-Body.
-`;
-            mockedExistsSync.mockImplementation((p) => {
-                const s = String(p);
-                return s.endsWith(".sweny/skills") || s.endsWith("sparse-config/SKILL.md");
-            });
-            mockedReaddirSync.mockReturnValue([{ name: "sparse-config", isDirectory: () => true }]);
-            mockedReadFileSync.mockReturnValue(skillMd);
-            const skills = discoverSkills("/fake");
-            expect(skills).toHaveLength(1);
-            // Falls back to key name as description
-            expect(skills[0].config.SOME_KEY.description).toBe("SOME_KEY");
-            expect(skills[0].config.SOME_KEY.required).toBe(true);
-            expect(skills[0].config.SOME_KEY.env).toBe("SOME_KEY");
-        });
-    });
-});

package/dist/__tests__/executor.test.d.ts

@@ -1 +0,0 @@
-export {};