opencrabs 0.3.44

The autonomous, self-improving AI agent. Single Rust binary. Every channel. Install with: cargo install opencrabs
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
//! Tests for the plan tool — security hardening + import operation.
//!
//! Originally lived inline at
//! `src/brain/tools/plan_tool_security_tests.rs` as a
//! `#[cfg(test)] mod tests { ... }` submodule of `plan_tool`. Moved
//! here as part of PR #160's review — the project convention is that
//! every test is a top-level file under `src/tests/` registered in
//! `tests/mod.rs`, no inline `#[cfg(test)] mod tests` blocks anywhere
//! else in the tree. Items the tests touch
//! (`validate_plan_file_path`, `validate_string`,
//! `MAX_PLAN_FILE_SIZE`, etc.) are now `pub(crate)` in `plan_tool.rs`
//! so this file can reach them from outside the module.

use crate::brain::tools::plan_tool::{
    MAX_CONTEXT_LENGTH, MAX_DESCRIPTION_LENGTH, MAX_PLAN_FILE_SIZE, MAX_TITLE_LENGTH, PlanTool,
    default_complexity, validate_plan_file_path, validate_string,
};
use crate::brain::tools::{Tool, ToolExecutionContext};
use std::path::PathBuf;
use tempfile::TempDir;

// ── path validation ───────────────────────────────────────────────

#[test]
fn validate_path_within_working_directory() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    let plan_file = working_dir.join(format!(".opencrabs_plan_{}.json", session_id));

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_ok());
}

#[test]
fn validate_path_outside_working_directory() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    // Try to write outside working directory
    let plan_file = PathBuf::from("/tmp").join(format!(".opencrabs_plan_{}.json", session_id));

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
    assert!(
        result
            .unwrap_err()
            .to_string()
            .contains("within the session directory")
    );
}

#[test]
fn validate_path_traversal_attack() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    // Try path traversal - construct a path that goes outside working_dir
    let parent = working_dir.parent().unwrap_or(working_dir);
    let plan_file = parent.join(format!(".opencrabs_plan_{}.json", session_id));

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
}

#[test]
fn validate_filename_pattern() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    // Invalid filename (not matching pattern)
    let plan_file = working_dir.join("invalid_plan.json");

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
    assert!(
        result
            .unwrap_err()
            .to_string()
            .contains("must match pattern")
    );
}

#[test]
fn validate_filename_requires_uuid() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    // Invalid UUID in filename
    let plan_file = working_dir.join(".opencrabs_plan_not-a-uuid.json");

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
    assert!(result.unwrap_err().to_string().contains("valid UUID"));
}

#[test]
#[cfg(unix)]
fn validate_symlink_rejection() {
    use std::os::unix::fs::symlink;

    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    let target_file = working_dir.join("target.json");
    let plan_file = working_dir.join(format!(".opencrabs_plan_{}.json", session_id));

    // Create a target file and symlink to it
    std::fs::write(&target_file, "{}").unwrap();
    symlink(&target_file, &plan_file).unwrap();

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
    assert!(result.unwrap_err().to_string().contains("symlink"));
}

// ── string validation ─────────────────────────────────────────────

#[test]
fn validate_string_empty() {
    let result = validate_string("", 100, "Test field");
    assert!(result.is_err());
    assert!(result.unwrap_err().to_string().contains("cannot be empty"));
}

#[test]
fn validate_string_whitespace_only() {
    let result = validate_string("   ", 100, "Test field");
    assert!(result.is_err());
    assert!(result.unwrap_err().to_string().contains("cannot be empty"));
}

#[test]
fn validate_string_exceeds_max_length() {
    let long_string = "a".repeat(300);
    let result = validate_string(&long_string, MAX_TITLE_LENGTH, "Title");
    assert!(result.is_err());
    assert!(
        result
            .unwrap_err()
            .to_string()
            .contains("exceeds maximum length")
    );
}

#[test]
fn validate_string_valid() {
    let result = validate_string("Valid title", MAX_TITLE_LENGTH, "Title");
    assert!(result.is_ok());
}

#[test]
fn max_plan_file_size_constant() {
    // Verify the constant is reasonable (10MB)
    assert_eq!(MAX_PLAN_FILE_SIZE, 10 * 1024 * 1024);
}

#[test]
fn input_validation_limits() {
    // Verify limits are reasonable
    assert_eq!(MAX_TITLE_LENGTH, 200);
    assert_eq!(MAX_DESCRIPTION_LENGTH, 5000);
    assert_eq!(MAX_CONTEXT_LENGTH, 5000);
}

#[test]
fn default_complexity_is_three() {
    assert_eq!(default_complexity(), 3);
}

#[test]
fn validate_title_at_limit() {
    let title = "a".repeat(MAX_TITLE_LENGTH);
    let result = validate_string(&title, MAX_TITLE_LENGTH, "Title");
    assert!(result.is_ok());
}

#[test]
fn validate_title_one_over_limit() {
    let title = "a".repeat(MAX_TITLE_LENGTH + 1);
    let result = validate_string(&title, MAX_TITLE_LENGTH, "Title");
    assert!(result.is_err());
}

#[test]
fn validate_description_at_limit() {
    let desc = "a".repeat(MAX_DESCRIPTION_LENGTH);
    let result = validate_string(&desc, MAX_DESCRIPTION_LENGTH, "Description");
    assert!(result.is_ok());
}

#[test]
fn validate_context_at_limit() {
    let context = "a".repeat(MAX_CONTEXT_LENGTH);
    let result = validate_string(&context, MAX_CONTEXT_LENGTH, "Context");
    assert!(result.is_ok());
}

#[test]
fn filename_with_special_characters() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    // Try filename with special characters that might be injection attempts
    let plan_file = working_dir.join(".opencrabs_plan_../../etc/passwd.json");

    let result = validate_plan_file_path(&plan_file, working_dir);
    assert!(result.is_err());
}

#[test]
fn filename_with_null_byte() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    let filename = format!(".opencrabs_plan_{}\0.json", session_id);
    let plan_file = working_dir.join(filename);

    // Rust's Path handling should prevent null bytes, but test anyway
    let result = validate_plan_file_path(&plan_file, working_dir);
    // Either fails validation or panic is caught
    assert!(result.is_err() || plan_file.to_str().is_none());
}

#[test]
fn validate_plan_file_path_canonical() {
    let temp_dir = TempDir::new().unwrap();
    let working_dir = temp_dir.path();

    let session_id = uuid::Uuid::new_v4();
    // Use ./ which should resolve to working_dir
    let plan_file = working_dir.join(format!("./.opencrabs_plan_{}.json", session_id));

    // Should still validate correctly after canonicalization
    let result = validate_plan_file_path(&plan_file, working_dir);
    // May pass or fail depending on path resolution, but shouldn't panic
    let _ = result;
}

// ── import operation ──────────────────────────────────────────────
//
// PR #160 added the import operation alongside the sample plan
// fixture. The tests below cover the happy path plus the four
// error / hardening paths the original PR was missing: size cap,
// invalid JSON, orphan dependency UUIDs, and symlink rejection at
// the target file.

#[tokio::test]
async fn import_sample_plan_succeeds() {
    let json = include_str!("../brain/tools/test_data/sample-coding-plan.json");

    let tmp_dir = TempDir::new().unwrap();
    let plan_file = tmp_dir.path().join("sample-coding-plan.json");
    std::fs::write(&plan_file, json).unwrap();

    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;

    let input = serde_json::json!({
        "operation": "import",
        "file_path": plan_file.to_str().unwrap(),
    });

    let result = tool.execute(input, &ctx).await.unwrap();
    assert!(result.success, "import must succeed on the sample plan");
    assert!(result.output.contains("Imported plan"));
    assert!(result.output.contains("7 tasks"));
}

#[tokio::test]
async fn import_rejects_file_over_size_cap() {
    // 10 MB + 1 byte triggers the size check before parse. This guards
    // against a malicious or runaway plan file blowing up memory on
    // read_to_string. The bytes don't need to be valid UTF-8 since the
    // size check fires before any parsing.
    let tmp_dir = TempDir::new().unwrap();
    let plan_file = tmp_dir.path().join("too_big.json");
    let payload = vec![b'a'; 10 * 1024 * 1024 + 1];
    std::fs::write(&plan_file, payload).unwrap();

    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;
    let input = serde_json::json!({
        "operation": "import",
        "file_path": plan_file.to_str().unwrap(),
    });

    let err = tool
        .execute(input, &ctx)
        .await
        .expect_err("oversize import must error");
    let msg = err.to_string();
    assert!(
        msg.contains("too large"),
        "expected 'too large' size-cap error, got: {msg}"
    );
}

#[tokio::test]
async fn import_rejects_invalid_json() {
    let tmp_dir = TempDir::new().unwrap();
    let plan_file = tmp_dir.path().join("bad.json");
    std::fs::write(&plan_file, "{this is not valid json").unwrap();

    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;
    let input = serde_json::json!({
        "operation": "import",
        "file_path": plan_file.to_str().unwrap(),
    });

    let err = tool
        .execute(input, &ctx)
        .await
        .expect_err("malformed JSON import must error");
    let msg = err.to_string();
    assert!(
        msg.contains("Invalid plan JSON"),
        "expected 'Invalid plan JSON' error, got: {msg}"
    );
}

#[tokio::test]
async fn import_rejects_orphan_dependency_uuid() {
    // A dependency that references a UUID not present in the imported
    // task set is a malformed plan. Silent `filter_map` dropping such
    // refs hid authoring mistakes; the import must reject with a
    // specific error so the user can fix the JSON.
    let bad_json = r#"{
        "id": "00000000-0000-0000-0000-000000000000",
        "session_id": "00000000-0000-0000-0000-000000000000",
        "title": "Bad Deps",
        "description": "Has a dep on a UUID not in the task list",
        "status": "Draft",
        "context": "",
        "risks": [],
        "test_strategy": "",
        "technical_stack": [],
        "created_at": "2026-01-01T00:00:00Z",
        "updated_at": "2026-01-01T00:00:00Z",
        "approved_at": null,
        "tasks": [
            {
                "id": "11111111-1111-1111-1111-111111111111",
                "order": 1,
                "title": "Orphan dep task",
                "description": "Depends on a uuid that isn't here",
                "task_type": "Edit",
                "dependencies": ["99999999-9999-9999-9999-999999999999"],
                "complexity": 1,
                "acceptance_criteria": [],
                "status": "Pending",
                "notes": null,
                "completed_at": null
            }
        ]
    }"#;

    let tmp_dir = TempDir::new().unwrap();
    let plan_file = tmp_dir.path().join("orphan_dep.json");
    std::fs::write(&plan_file, bad_json).unwrap();

    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;
    let input = serde_json::json!({
        "operation": "import",
        "file_path": plan_file.to_str().unwrap(),
    });

    let err = tool
        .execute(input, &ctx)
        .await
        .expect_err("orphan-dep import must error");
    let msg = err.to_string();
    assert!(
        msg.contains("depends on unknown task id"),
        "expected orphan-dep error, got: {msg}"
    );
}

#[tokio::test]
#[cfg(unix)]
async fn import_rejects_symlink_at_target() {
    // The symlink check on the TARGET file (the import file itself)
    // still has to fire — a malicious user could place a symlink at
    // the import location pointing somewhere else and trick the agent
    // into reading from the resolved target. The PR's original
    // ancestor-walking approach was wrong (broke on macOS where /var
    // is a symlink), but the target-only check still has to catch a
    // symlink at the file itself.
    let tmp_dir = TempDir::new().unwrap();
    let real_file = tmp_dir.path().join("real.json");
    std::fs::write(&real_file, "{}").unwrap();
    let symlink_path = tmp_dir.path().join("link.json");
    std::os::unix::fs::symlink(&real_file, &symlink_path).unwrap();

    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;
    let input = serde_json::json!({
        "operation": "import",
        "file_path": symlink_path.to_str().unwrap(),
    });

    let err = tool
        .execute(input, &ctx)
        .await
        .expect_err("symlink target import must error");
    let msg = err.to_string();
    assert!(
        msg.contains("symlink"),
        "expected symlink rejection, got: {msg}"
    );
}

// ── start_task auto-completion ─────────────────────────────────────

#[tokio::test]
async fn start_task_auto_completes_previous_in_progress_task() {
    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;

    // Create a plan with 3 tasks
    tool.execute(
        serde_json::json!({
            "operation": "create",
            "title": "Auto-complete test",
            "description": "Testing auto-completion of previous task"
        }),
        &ctx,
    )
    .await
    .unwrap();

    for i in 1..=3 {
        tool.execute(
            serde_json::json!({
                "operation": "add_task",
                "title": format!("Task {}", i),
                "description": format!("Description {}", i),
                "task_type": "edit"
            }),
            &ctx,
        )
        .await
        .unwrap();
    }

    tool.execute(serde_json::json!({ "operation": "finalize" }), &ctx)
        .await
        .unwrap();

    // Start task 1
    tool.execute(
        serde_json::json!({ "operation": "start_task", "task_order": 1 }),
        &ctx,
    )
    .await
    .unwrap();

    // Start task 2 WITHOUT completing task 1 first.
    // This should auto-complete task 1.
    let result = tool
        .execute(
            serde_json::json!({ "operation": "start_task", "task_order": 2 }),
            &ctx,
        )
        .await
        .unwrap();

    assert!(
        result.output.contains("Auto-completed"),
        "should auto-complete previous task, got: {}",
        result.output
    );
    assert!(
        result.output.contains("Task 1"),
        "should mention the auto-completed task name, got: {}",
        result.output
    );

    // Verify task 1 is now completed via summary
    let summary = tool
        .execute(serde_json::json!({ "operation": "summary" }), &ctx)
        .await
        .unwrap();
    assert!(
        summary.output.contains("[x] Task 1"),
        "task 1 should be completed, got: {}",
        summary.output
    );
    assert!(
        summary.output.contains("[>] Task 2"),
        "task 2 should be in progress, got: {}",
        summary.output
    );
}

#[tokio::test]
async fn start_task_no_auto_complete_when_no_previous_in_progress() {
    let ctx = ToolExecutionContext::new(uuid::Uuid::new_v4());
    let tool = PlanTool;

    tool.execute(
        serde_json::json!({
            "operation": "create",
            "title": "No auto-complete test",
            "description": "Normal flow"
        }),
        &ctx,
    )
    .await
    .unwrap();

    for i in 1..=2 {
        tool.execute(
            serde_json::json!({
                "operation": "add_task",
                "title": format!("Task {}", i),
                "description": format!("Desc {}", i),
                "task_type": "edit"
            }),
            &ctx,
        )
        .await
        .unwrap();
    }

    tool.execute(serde_json::json!({ "operation": "finalize" }), &ctx)
        .await
        .unwrap();

    // Start task 1, complete it properly
    tool.execute(
        serde_json::json!({ "operation": "start_task", "task_order": 1 }),
        &ctx,
    )
    .await
    .unwrap();

    tool.execute(
        serde_json::json!({
            "operation": "complete_task",
            "task_order": 1,
            "success": true,
            "output": "Done"
        }),
        &ctx,
    )
    .await
    .unwrap();

    // Start task 2 - no auto-complete should happen since task 1 is already completed
    let result = tool
        .execute(
            serde_json::json!({ "operation": "start_task", "task_order": 2 }),
            &ctx,
        )
        .await
        .unwrap();

    assert!(
        !result.output.contains("Auto-completed"),
        "should NOT auto-complete when previous task is already done, got: {}",
        result.output
    );
}