cognitive-core 0.2.4 → 0.3.1

package/tests/search/verification-runner.test.ts

@@ -1,357 +0,0 @@
-import { describe, it, expect, beforeEach } from 'vitest';
-import {
-  VerificationRunner,
-  createVerificationRunner,
-  TestRunners,
-  type CommandVerificationConfig,
-} from '../../src/search/verification-runner.js';
-import { createTrajectory } from '../../src/types/trajectory.js';
-import { createTask } from '../../src/types/task.js';
-import { createStep } from '../../src/types/step.js';
-import { successOutcome } from '../../src/types/outcome.js';
-import type { Trajectory } from '../../src/types/trajectory.js';
-import type { Task } from '../../src/types/task.js';
-
-function makeTask(): Task {
-  return createTask({
-    domain: 'test-domain',
-    description: 'test task',
-  });
-}
-
-function makeTrajectory(): Trajectory {
-  return createTrajectory({
-    task: makeTask(),
-    steps: [createStep({ action: 'echo hello', observation: 'hello' })],
-    outcome: successOutcome('hello world'),
-    agentId: 'agent',
-  });
-}
-
-describe('VerificationRunner', () => {
-  let runner: VerificationRunner;
-
-  beforeEach(() => {
-    runner = createVerificationRunner({ cwd: '/tmp', timeout: 5000 });
-  });
-
-  describe('runCommand', () => {
-    it('should execute a simple command and capture stdout', async () => {
-      const result = await runner.runCommand('echo hello', {});
-      expect(result.stdout.trim()).toBe('hello');
-      expect(result.exitCode).toBe(0);
-      expect(result.timedOut).toBe(false);
-    });
-
-    it('should capture stderr', async () => {
-      const result = await runner.runCommand('echo error >&2', {});
-      expect(result.stderr.trim()).toBe('error');
-    });
-
-    it('should return non-zero exit code for failing commands', async () => {
-      const result = await runner.runCommand('exit 42', {});
-      expect(result.exitCode).toBe(42);
-    });
-
-    it('should handle command timeout', { timeout: 20000 }, async () => {
-      // Use a bash loop that responds to SIGTERM better than plain sleep
-      const result = await runner.runCommand('while true; do sleep 0.1; done', { timeout: 500 });
-      expect(result.timedOut).toBe(true);
-    });
-
-    it('should handle command not found', async () => {
-      const result = await runner.runCommand('nonexistent_command_xyz_12345', {});
-      expect(result.exitCode).not.toBe(0);
-    });
-
-    it('should pass environment variables', async () => {
-      const result = await runner.runCommand('echo $MY_TEST_VAR', {
-        env: { MY_TEST_VAR: 'hello123' },
-      });
-      expect(result.stdout.trim()).toBe('hello123');
-    });
-
-    it('should use specified working directory', async () => {
-      const result = await runner.runCommand('pwd', { cwd: '/tmp' });
-      expect(result.stdout.trim()).toMatch(/\/tmp/);
-    });
-  });
-
-  describe('runCommandSync', () => {
-    it('should execute command synchronously', () => {
-      const result = runner.runCommandSync('echo sync-test');
-      expect(result.stdout.trim()).toBe('sync-test');
-      expect(result.exitCode).toBe(0);
-    });
-
-    it('should handle sync command failure', () => {
-      const result = runner.runCommandSync('exit 1');
-      expect(result.exitCode).toBe(1);
-    });
-
-    it('should detect sync command failure on timeout', () => {
-      const result = runner.runCommandSync('sleep 30', { timeout: 500 });
-      // execSync throws on timeout; killed + SIGTERM detection may vary by platform
-      // The important thing is that the command does not succeed
-      expect(result.exitCode).not.toBe(0);
-    });
-  });
-
-  describe('createVerifier', () => {
-    it('should create verifier that passes on exit code 0', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "all tests pass"',
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.passed).toBe(true);
-      expect(result.confidence).toBe(0.7); // No pattern/extractor → 0.7
-    });
-
-    it('should create verifier that fails on non-zero exit', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'exit 1',
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.passed).toBe(false);
-      expect(result.issues!.length).toBeGreaterThan(0);
-    });
-
-    it('should check success pattern when provided', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "Tests: 5 passed, 0 failed"',
-        successPattern: /\d+ passed, 0 failed/,
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.passed).toBe(true);
-      expect(result.confidence).toBe(0.9); // Has successPattern → 0.9
-    });
-
-    it('should check failure pattern when provided', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "FAIL test_login"',
-        failurePattern: /FAIL/,
-        successOnZeroExit: false,
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.passed).toBe(false);
-    });
-
-    it('should use custom issue extractor', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "ERROR: missing return"',
-        timeout: 5000,
-        successOnZeroExit: false,
-        issueExtractor: (output) => {
-          const match = output.match(/ERROR:\s*(.+)/);
-          return match
-            ? [{ type: 'error' as const, description: match[1], severity: 'critical' as const }]
-            : [];
-        },
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.confidence).toBe(0.9); // Has issueExtractor
-    });
-
-    it('should handle timeout in verifier', { timeout: 20000 }, async () => {
-      const config: CommandVerificationConfig = {
-        command: 'while true; do sleep 0.1; done',
-        timeout: 500,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.passed).toBe(false);
-      expect(result.confidence).toBe(0.5); // Timeout → 0.5
-      expect(result.details).toContain('timed out');
-    });
-
-    it('should interpolate {{taskId}} and {{trajectoryId}} in command', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "task={{taskId}} traj={{trajectoryId}}"',
-        timeout: 5000,
-      };
-
-      const trajectory = makeTrajectory();
-      const task = makeTask();
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(trajectory, task);
-
-      expect(result.details).toContain(`task=${task.id}`);
-      expect(result.details).toContain(`traj=${trajectory.id}`);
-    });
-  });
-
-  describe('createCompositeVerifier', () => {
-    it('should pass when all verifiers pass', async () => {
-      const composite = runner.createCompositeVerifier([
-        { command: 'echo pass1', timeout: 5000 },
-        { command: 'echo pass2', timeout: 5000 },
-      ]);
-
-      const result = await composite(makeTrajectory(), makeTask());
-      expect(result.passed).toBe(true);
-    });
-
-    it('should fail when any verifier fails', async () => {
-      const composite = runner.createCompositeVerifier([
-        { command: 'echo pass', timeout: 5000 },
-        { command: 'exit 1', timeout: 5000 },
-      ]);
-
-      const result = await composite(makeTrajectory(), makeTask());
-      expect(result.passed).toBe(false);
-    });
-
-    it('should aggregate issues from all verifiers', async () => {
-      const composite = runner.createCompositeVerifier([
-        { command: 'exit 1', timeout: 5000 },
-        { command: 'exit 2', timeout: 5000 },
-      ]);
-
-      const result = await composite(makeTrajectory(), makeTask());
-      expect(result.issues!.length).toBeGreaterThanOrEqual(1);
-    });
-
-    it('should fail fast on critical issues', async () => {
-      let secondRan = false;
-      const configs: CommandVerificationConfig[] = [
-        {
-          command: 'exit 1',
-          timeout: 5000,
-          issueExtractor: () => [
-            { type: 'error', description: 'critical failure', severity: 'critical' },
-          ],
-        },
-        {
-          command: 'echo should-not-run',
-          timeout: 5000,
-          issueExtractor: () => {
-            secondRan = true;
-            return [];
-          },
-        },
-      ];
-
-      const composite = runner.createCompositeVerifier(configs);
-      await composite(makeTrajectory(), makeTask());
-
-      // Second verifier should not have run due to fail-fast
-      expect(secondRan).toBe(false);
-    });
-
-    it('should average confidence across verifiers', async () => {
-      const composite = runner.createCompositeVerifier([
-        { command: 'echo pass', timeout: 5000 }, // confidence 0.7
-        { command: 'echo pass', timeout: 5000 }, // confidence 0.7
-      ]);
-
-      const result = await composite(makeTrajectory(), makeTask());
-      expect(result.confidence).toBeCloseTo(0.7, 1);
-    });
-  });
-
-  describe('TestRunners presets', () => {
-    it('should create vitest configuration', () => {
-      const config = TestRunners.vitest('tests/foo.test.ts');
-      expect(config.command).toContain('vitest');
-      expect(config.command).toContain('tests/foo.test.ts');
-      expect(config.timeout).toBe(60000);
-      expect(config.issueExtractor).toBeDefined();
-    });
-
-    it('should create jest configuration', () => {
-      const config = TestRunners.jest();
-      expect(config.command).toContain('jest');
-      expect(config.issueExtractor).toBeDefined();
-    });
-
-    it('should create typescript configuration', () => {
-      const config = TestRunners.typescript();
-      expect(config.command).toContain('tsc --noEmit');
-      expect(config.issueExtractor).toBeDefined();
-    });
-
-    it('should create eslint configuration', () => {
-      const config = TestRunners.eslint('src/');
-      expect(config.command).toContain('eslint');
-      expect(config.command).toContain('src/');
-    });
-
-    it('should create pytest configuration', () => {
-      const config = TestRunners.pytest();
-      expect(config.command).toContain('pytest');
-      expect(config.failurePattern).toBeDefined();
-    });
-
-    it('should create go test configuration', () => {
-      const config = TestRunners.goTest();
-      expect(config.command).toContain('go test');
-    });
-
-    it('should create cargo test configuration', () => {
-      const config = TestRunners.cargoTest();
-      expect(config.command).toContain('cargo test');
-      expect(config.timeout).toBe(120000);
-    });
-  });
-
-  describe('default issue extraction', () => {
-    it('should extract error messages from output', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "Error: cannot read file" && exit 1',
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.issues!.length).toBeGreaterThan(0);
-      expect(result.issues![0].description).toContain('cannot read file');
-    });
-
-    it('should extract fail messages from output', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'echo "Fail: test_login" && exit 1',
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.issues!.length).toBeGreaterThan(0);
-    });
-
-    it('should generate generic issue when no pattern matches', async () => {
-      const config: CommandVerificationConfig = {
-        command: 'exit 99',
-        timeout: 5000,
-      };
-
-      const verifier = runner.createVerifier(config);
-      const result = await verifier(makeTrajectory(), makeTask());
-
-      expect(result.issues!.length).toBeGreaterThan(0);
-      expect(result.issues![0].description).toContain('exit code 99');
-    });
-  });
-});

package/tests/session-bank/fixtures/multi-tool-session/full.jsonl

@@ -1,12 +0,0 @@
-{"type":"user","uuid":"msg-101","message":{"content":"Add unit tests for the utils module and fix any bugs you find"}}
-{"type":"assistant","uuid":"msg-102","message":{"content":[{"type":"text","text":"Let me start by reading the utils module to understand what needs testing."},{"type":"tool_use","id":"tool-101","name":"Read","input":{"file_path":"/src/utils.ts"}}]}}
-{"type":"user","uuid":"msg-103","message":{"content":[{"type":"tool_result","tool_use_id":"tool-101","content":"export function slugify(s: string): string {\n  return s.toLowerCase().replace(/[^a-z0-9]+/g, '-');\n}\n\nexport function truncate(s: string, len: number): string {\n  if (s.length <= len) return s;\n  return s.slice(0, len);\n}"}]}}
-{"type":"assistant","uuid":"msg-104","message":{"content":[{"type":"text","text":"I see two functions. The `truncate` function doesn't add an ellipsis — that might be intentional but let me also check if there are existing tests."},{"type":"tool_use","id":"tool-102","name":"Glob","input":{"pattern":"tests/**/*.test.ts"}}]}}
-{"type":"user","uuid":"msg-105","message":{"content":[{"type":"tool_result","tool_use_id":"tool-102","content":"tests/auth.test.ts"}]}}
-{"type":"assistant","uuid":"msg-106","message":{"content":[{"type":"text","text":"No existing utils tests. Let me create them."},{"type":"tool_use","id":"tool-103","name":"Write","input":{"file_path":"/tests/utils.test.ts","content":"import { slugify, truncate } from '../src/utils';\nimport { describe, it, expect } from 'vitest';\n\ndescribe('slugify', () => {\n  it('lowercases and replaces non-alphanumeric', () => {\n    expect(slugify('Hello World!')).toBe('hello-world-');\n  });\n});\n\ndescribe('truncate', () => {\n  it('truncates long strings', () => {\n    expect(truncate('hello world', 5)).toBe('hello');\n  });\n  it('returns short strings unchanged', () => {\n    expect(truncate('hi', 5)).toBe('hi');\n  });\n});"}}]}}
-{"type":"user","uuid":"msg-107","message":{"content":[{"type":"tool_result","tool_use_id":"tool-103","content":"File created successfully"}]}}
-{"type":"assistant","uuid":"msg-108","message":{"content":[{"type":"text","text":"Now let me run the tests to make sure they pass."},{"type":"tool_use","id":"tool-104","name":"Bash","input":{"command":"npx vitest run tests/utils.test.ts"}}]}}
-{"type":"user","uuid":"msg-109","message":{"content":[{"type":"tool_result","tool_use_id":"tool-104","content":"✓ slugify > lowercases and replaces non-alphanumeric\n✓ truncate > truncates long strings\n✓ truncate > returns short strings unchanged\n\nTest Files  1 passed\nTests  3 passed"}]}}
-{"type":"assistant","uuid":"msg-110","message":{"content":[{"type":"text","text":"All 3 tests pass. I also noticed the `slugify` function leaves a trailing dash for trailing non-alphanumeric characters. Let me also check the Python helper."},{"type":"tool_use","id":"tool-105","name":"Read","input":{"file_path":"/src/helpers.py"}},{"type":"tool_use","id":"tool-106","name":"Bash","input":{"command":"python -m pytest tests/ -q"}}]}}
-{"type":"user","uuid":"msg-111","message":{"content":[{"type":"tool_result","tool_use_id":"tool-105","content":"def parse_config(path):\n    with open(path) as f:\n        return json.load(f)"},{"type":"tool_result","tool_use_id":"tool-106","content":"no tests ran","is_error":false}]}}
-{"type":"assistant","uuid":"msg-112","message":{"content":[{"type":"text","text":"Done. I created unit tests for the utils module — all 3 tests pass. The Python helper has a simple config parser but no tests yet. The slugify trailing-dash behavior could be improved in a follow-up."}]}}

package/tests/session-bank/fixtures/multi-tool-session/metadata.json

@@ -1,28 +0,0 @@
-{
-  "cli_version": "0.3.0",
-  "checkpoint_id": "f10a9b8c7d6e",
-  "session_id": "2026-01-09-def456ab-c789-0123-def4-567890abcdef",
-  "strategy": "auto-commit",
-  "created_at": "2026-01-09T14:15:00Z",
-  "branch": "feature/add-tests",
-  "files_touched": ["src/utils.ts", "tests/utils.test.ts", "src/helpers.py"],
-  "agent": "Claude Code",
-  "is_task": false,
-  "checkpoint_transcript_start": 0,
-  "token_usage": {
-    "input_tokens": 42000,
-    "cache_creation_tokens": 8000,
-    "cache_read_tokens": 12000,
-    "output_tokens": 9500,
-    "api_call_count": 15,
-    "subagent_usage": [
-      {
-        "input_tokens": 3000,
-        "cache_creation_tokens": 0,
-        "cache_read_tokens": 0,
-        "output_tokens": 1200,
-        "api_call_count": 2
-      }
-    ]
-  }
-}

package/tests/session-bank/fixtures/root-metadata.json

@@ -1,18 +0,0 @@
-{
-  "sessions": [
-    {
-      "metadata": "0/metadata.json",
-      "transcript": "0/full.jsonl",
-      "context": "0/context.md",
-      "content_hash": "0/content_hash.txt",
-      "prompt": "0/prompt.txt"
-    },
-    {
-      "metadata": "1/metadata.json",
-      "transcript": "1/full.jsonl",
-      "context": "1/context.md",
-      "content_hash": "1/content_hash.txt",
-      "prompt": "1/prompt.txt"
-    }
-  ]
-}

package/tests/session-bank/fixtures/sessionlog-root-metadata.json

@@ -1,16 +0,0 @@
-{
-  "cliVersion": "opentasks-sessionlog",
-  "checkpointID": "c4d5e6f7a8b9",
-  "strategy": "manual-commit",
-  "branch": "main",
-  "checkpointsCount": 3,
-  "filesTouched": ["src/api.ts", "tests/api.test.ts"],
-  "sessions": [
-    {
-      "metadata": "1/metadata.json",
-      "transcript": "1/full.jsonl",
-      "context": "1/context.md",
-      "prompt": "1/prompt.txt"
-    }
-  ]
-}

package/tests/session-bank/fixtures/sessionlog-session/full.jsonl

@@ -1,6 +0,0 @@
-{"type":"user","uuid":"msg-sl-001","message":{"content":"Add input validation to the API endpoint using Zod"}}
-{"type":"assistant","uuid":"msg-sl-002","message":{"content":[{"type":"text","text":"I'll examine the API endpoint first."},{"type":"tool_use","id":"tool-sl-001","name":"Read","input":{"file_path":"/src/api.ts"}}]}}
-{"type":"user","uuid":"msg-sl-003","message":{"content":[{"type":"tool_result","tool_use_id":"tool-sl-001","content":"export function handler(req, res) {\n  const body = req.body;\n  // TODO: validate\n  res.json({ ok: true });\n}"}]}}
-{"type":"assistant","uuid":"msg-sl-004","message":{"content":[{"type":"text","text":"I'll add Zod validation."},{"type":"tool_use","id":"tool-sl-002","name":"Edit","input":{"file_path":"/src/api.ts","old_string":"const body = req.body;","new_string":"const body = schema.parse(req.body);"}}]}}
-{"type":"user","uuid":"msg-sl-005","message":{"content":[{"type":"tool_result","tool_use_id":"tool-sl-002","content":"File edited successfully"}]}}
-{"type":"assistant","uuid":"msg-sl-006","message":{"content":[{"type":"text","text":"Validation added. The endpoint now validates input using the Zod schema before processing."}]}}

package/tests/session-bank/fixtures/sessionlog-session/metadata.json

@@ -1,55 +0,0 @@
-{
-  "cliVersion": "opentasks-sessionlog",
-  "checkpointID": "c4d5e6f7a8b9",
-  "sessionID": "2026-02-15-aabbccdd-1122-3344-5566-778899aabbcc",
-  "strategy": "manual-commit",
-  "createdAt": "2026-02-15T09:00:00Z",
-  "branch": "main",
-  "filesTouched": ["src/api.ts", "tests/api.test.ts"],
-  "agent": "Claude Code",
-  "turnID": "turn-001",
-  "isTask": false,
-  "checkpointTranscriptStart": 0,
-  "tokenUsage": {
-    "inputTokens": 20000,
-    "cacheCreationTokens": 3000,
-    "cacheReadTokens": 6000,
-    "outputTokens": 4000,
-    "apiCallCount": 10,
-    "subagentTokens": {
-      "inputTokens": 2000,
-      "cacheCreationTokens": 0,
-      "cacheReadTokens": 0,
-      "outputTokens": 800,
-      "apiCallCount": 3
-    }
-  },
-  "summary": {
-    "intent": "Add validation to the API endpoint",
-    "outcome": "Added input validation with Zod schema",
-    "learnings": {
-      "repo": ["API routes use Zod for validation"],
-      "code": [
-        {
-          "path": "src/api.ts",
-          "line": 15,
-          "finding": "Zod schema must be defined before route handler"
-        }
-      ],
-      "workflow": ["Run API tests with --reporter=verbose flag"]
-    },
-    "friction": ["Had to look up Zod coercion syntax"],
-    "openItems": ["Add validation for the PUT endpoint too"]
-  },
-  "initialAttribution": {
-    "calculatedAt": "2026-02-15T09:30:00Z",
-    "agentLines": 30,
-    "humanAdded": 3,
-    "humanModified": 1,
-    "humanRemoved": 0,
-    "totalCommitted": 34,
-    "agentPercentage": 0.88
-  },
-  "tasks": {},
-  "skillsUsed": []
-}

package/tests/session-bank/fixtures/simple-session/full.jsonl

@@ -1,6 +0,0 @@
-{"type":"user","uuid":"msg-001","message":{"content":"Fix the bug in auth.ts where the token refresh doesn't work"}}
-{"type":"assistant","uuid":"msg-002","message":{"content":[{"type":"text","text":"I'll look at the auth.ts file to understand the token refresh logic."},{"type":"tool_use","id":"tool-001","name":"Read","input":{"file_path":"/src/auth.ts"}}]}}
-{"type":"user","uuid":"msg-003","message":{"content":[{"type":"tool_result","tool_use_id":"tool-001","content":"export async function refreshToken() {\n  const token = tokenCache.get();\n  fetchNewToken(token);\n  return token;\n}"}]}}
-{"type":"assistant","uuid":"msg-004","message":{"content":[{"type":"text","text":"I found the issue — `fetchNewToken` is async but not awaited. Let me fix it."},{"type":"tool_use","id":"tool-002","name":"Edit","input":{"file_path":"/src/auth.ts","old_string":"fetchNewToken(token);","new_string":"await fetchNewToken(token);"}}]}}
-{"type":"user","uuid":"msg-005","message":{"content":[{"type":"tool_result","tool_use_id":"tool-002","content":"File edited successfully"}]}}
-{"type":"assistant","uuid":"msg-006","message":{"content":[{"type":"text","text":"The fix is applied. The `fetchNewToken` call was missing `await`, which meant the token wasn't actually refreshed before being returned."}]}}

package/tests/session-bank/fixtures/simple-session/metadata.json

@@ -1,38 +0,0 @@
-{
-  "cli_version": "0.3.0",
-  "checkpoint_id": "a3b2c4d5e6f7",
-  "session_id": "2026-01-08-abc123de-f456-7890-abcd-ef1234567890",
-  "strategy": "manual-commit",
-  "created_at": "2026-01-08T10:30:00Z",
-  "branch": "main",
-  "files_touched": ["src/auth.ts", "src/middleware.ts"],
-  "agent": "Claude Code",
-  "is_task": false,
-  "checkpoint_transcript_start": 0,
-  "token_usage": {
-    "input_tokens": 15000,
-    "cache_creation_tokens": 2000,
-    "cache_read_tokens": 5000,
-    "output_tokens": 3000,
-    "api_call_count": 8
-  },
-  "summary": {
-    "intent": "Fix the authentication token refresh bug in the middleware",
-    "outcome": "Successfully fixed the token refresh logic and added retry handling",
-    "learnings": {
-      "repo_patterns": ["Auth middleware uses a singleton token cache"],
-      "code_findings": ["Token refresh was not awaiting the async call"],
-      "workflow_practices": ["Run auth tests before committing middleware changes"]
-    },
-    "friction_points": ["Initial confusion about which middleware file to edit"],
-    "open_items": ["Add integration test for token expiry edge case"]
-  },
-  "initial_attribution": {
-    "agent_lines_added": 25,
-    "human_additions": 2,
-    "human_modifications": 1,
-    "human_removals": 0,
-    "total_committed_lines": 28,
-    "agent_contribution": 0.89
-  }
-}