agentics-contracts 0.4.0

Shared API and challenge contract DTOs for Agentics.
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
use super::*;

#[tokio::test]
async fn disabled_private_benchmark_bundle_does_not_require_directory() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-disabled-private-benchmark-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    create_bundle(&root, &spec);

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    assert!(result.is_ok());
}

/// Verifies that source backed run inputs must exist under bundle root.
#[tokio::test]
async fn source_backed_run_inputs_must_exist_under_bundle_root() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-source-input-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    std::fs::write(
        root.join("public/runs.json"),
        r#"{"runs":[{"run_name":"public-1","interface":"file_system","stdin_json":null,"stdin_text":null,"input_files":[{"path":"input.txt","source_path":"public/input.txt"}],"output_files":["answer.txt"],"metadata":null}]}"#,
    )
    .expect("failed to write source-backed runs");

    let missing_result = validate_challenge_bundle(&root).await;
    std::fs::write(root.join("public/input.txt"), "payload\n")
        .expect("failed to write source input");
    let present_result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    assert!(missing_result.is_err());
    assert!(present_result.is_ok());
}

/// Verifies source-backed session inputs are validated under the selected source root.
#[tokio::test]
async fn source_backed_session_inputs_must_exist_under_bundle_root() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-source-session-input-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_piped_stdio_spec();
    spec.datasets.private_benchmark_enabled = false;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    if let ChallengeExecutionSpec::PipedStdio(execution) = &mut spec.execution {
        execution.official_session = None;
    }
    create_piped_stdio_bundle(&root, &spec);
    std::fs::remove_file(root.join("public/prompt.txt")).expect("failed to remove source input");

    let missing_result = validate_challenge_bundle(&root).await;
    std::fs::write(root.join("public/prompt.txt"), "payload\n")
        .expect("failed to restore source input");
    let present_result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    assert!(missing_result.is_err());
    assert!(present_result.is_ok());
}

/// Verifies session manifests reject duplicate materialized input paths.
#[tokio::test]
async fn session_manifest_rejects_duplicate_input_paths() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-duplicate-session-input-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_piped_stdio_spec();
    spec.datasets.private_benchmark_enabled = false;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    if let ChallengeExecutionSpec::PipedStdio(execution) = &mut spec.execution {
        execution.official_session = None;
    }
    create_piped_stdio_bundle(&root, &spec);
    std::fs::write(
        root.join("public/session.json"),
        r#"{"session_name":"public-1","input_files":[{"path":"prompt.txt","content":"a"},{"path":"prompt.txt","content":"b"}],"metadata":null}"#,
    )
    .expect("failed to write duplicate session inputs");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("duplicate session input paths should fail");
    assert!(error.to_string().contains("duplicate path"));
}

/// Verifies separated-evaluator run manifests reject duplicate materialized input paths.
#[tokio::test]
async fn run_manifest_rejects_duplicate_input_paths() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-duplicate-run-input-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    std::fs::write(
        root.join("public/runs.json"),
        r#"{"runs":[{"run_name":"public-1","interface":"file_system","stdin_json":null,"stdin_text":null,"input_files":[{"path":"prompt.txt","content":"a"},{"path":"prompt.txt","content":"b"}],"output_files":null,"metadata":null}]}"#,
    )
    .expect("failed to write duplicate run inputs");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("duplicate run input paths should fail");
    assert!(error.to_string().contains("duplicate path"));
}

/// Verifies separated-evaluator run manifests require nullable fields to be present.
#[tokio::test]
async fn run_manifest_requires_nullable_fields_to_be_present() {
    for field in [
        "stdin_json",
        "stdin_text",
        "input_files",
        "output_files",
        "metadata",
    ] {
        let root = std::env::temp_dir().join(format!(
            "agentics-bundle-missing-run-field-{field}-{}",
            uuid::Uuid::new_v4()
        ));
        let mut spec = base_spec();
        spec.datasets.private_benchmark_enabled = false;
        create_bundle(&root, &spec);
        let mut run = serde_json::json!({
            "run_name": "public-1",
            "interface": "stdio",
            "stdin_json": null,
            "stdin_text": "1",
            "input_files": null,
            "output_files": null,
            "metadata": null
        });
        run.as_object_mut()
            .expect("run should be an object")
            .remove(field);
        std::fs::write(
            root.join("public/runs.json"),
            serde_json::json!({ "runs": [run] }).to_string(),
        )
        .expect("failed to write incomplete run manifest");

        let result = validate_challenge_bundle(&root).await;
        drop(std::fs::remove_dir_all(root));

        let error = result.expect_err("missing required nullable run field should fail");
        assert!(
            error.to_string().contains(field),
            "unexpected error for {field}: {error}"
        );
    }
}

/// Verifies run manifests use null, not empty arrays, for absent input or output declarations.
#[tokio::test]
async fn run_manifest_rejects_empty_nullable_arrays() {
    for (field, message) in [
        ("input_files", "empty arrays must use null"),
        ("output_files", "empty arrays must use null"),
    ] {
        let root = std::env::temp_dir().join(format!(
            "agentics-bundle-empty-run-array-{field}-{}",
            uuid::Uuid::new_v4()
        ));
        let mut spec = base_spec();
        spec.datasets.private_benchmark_enabled = false;
        create_bundle(&root, &spec);
        let mut run = serde_json::json!({
            "run_name": "public-1",
            "interface": "stdio",
            "stdin_json": null,
            "stdin_text": "1",
            "input_files": null,
            "output_files": null,
            "metadata": null
        });
        run[field] = serde_json::json!([]);
        std::fs::write(
            root.join("public/runs.json"),
            serde_json::json!({ "runs": [run] }).to_string(),
        )
        .expect("failed to write run manifest with empty array");

        let result = validate_challenge_bundle(&root).await;
        drop(std::fs::remove_dir_all(root));

        let error = result.expect_err("empty nullable run arrays should fail");
        assert!(
            error.to_string().contains(message) || error.to_string().contains("non-empty array"),
            "unexpected error for {field}: {error}"
        );
    }
}

/// Verifies run metadata cannot shadow evaluator-visible reserved run fields.
#[tokio::test]
async fn run_manifest_metadata_cannot_shadow_reserved_fields() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-reserved-run-metadata-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    std::fs::write(
        root.join("public/runs.json"),
        r#"{"runs":[{"run_name":"public-1","interface":"stdio","stdin_json":null,"stdin_text":"1","input_files":null,"output_files":null,"metadata":{"run_name":"shadow"}}]}"#,
    )
    .expect("failed to write reserved run metadata");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("reserved metadata key should fail");
    assert!(error.to_string().contains("run_name"));
}

/// Verifies run names cannot collide with the runner-generated evaluator manifest.
#[tokio::test]
async fn run_manifest_rejects_generated_evaluator_manifest_name() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-reserved-run-name-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    std::fs::write(
        root.join("public/runs.json"),
        r#"{"runs":[{"run_name":"agentics-runs.json","interface":"stdio","stdin_json":null,"stdin_text":"1","input_files":null,"output_files":null,"metadata":null}]}"#,
    )
    .expect("failed to write reserved run name");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("reserved run_name should fail");
    assert!(error.to_string().contains("runner metadata"));
}

/// Verifies session metadata must be an object when present.
#[tokio::test]
async fn session_manifest_rejects_non_object_metadata() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-session-metadata-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_piped_stdio_spec();
    spec.datasets.private_benchmark_enabled = false;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    if let ChallengeExecutionSpec::PipedStdio(execution) = &mut spec.execution {
        execution.official_session = None;
    }
    create_piped_stdio_bundle(&root, &spec);
    std::fs::write(
        root.join("public/session.json"),
        r#"{"session_name":"public-1","input_files":null,"metadata":["not","object"]}"#,
    )
    .expect("failed to write invalid session metadata");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("non-object metadata should fail");
    assert!(error.to_string().contains("invalid session manifest"));
}

/// Verifies piped-stdio session manifests require nullable fields to be present.
#[tokio::test]
async fn session_manifest_requires_nullable_fields_to_be_present() {
    for field in ["input_files", "metadata"] {
        let root = std::env::temp_dir().join(format!(
            "agentics-bundle-missing-session-field-{field}-{}",
            uuid::Uuid::new_v4()
        ));
        let mut spec = base_piped_stdio_spec();
        spec.datasets.private_benchmark_enabled = false;
        spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
        if let ChallengeExecutionSpec::PipedStdio(execution) = &mut spec.execution {
            execution.official_session = None;
        }
        create_piped_stdio_bundle(&root, &spec);
        let mut session = serde_json::json!({
            "session_name": "public-1",
            "input_files": null,
            "metadata": null
        });
        session
            .as_object_mut()
            .expect("session should be an object")
            .remove(field);
        std::fs::write(root.join("public/session.json"), session.to_string())
            .expect("failed to write incomplete session manifest");

        let result = validate_challenge_bundle(&root).await;
        drop(std::fs::remove_dir_all(root));

        let error = result.expect_err("missing required nullable session field should fail");
        assert!(
            error.to_string().contains("invalid session manifest"),
            "unexpected error for {field}: {error}"
        );
    }
}

/// Verifies session manifests use null, not an empty array, for absent input files.
#[tokio::test]
async fn session_manifest_rejects_empty_input_file_array() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-empty-session-inputs-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_piped_stdio_spec();
    spec.datasets.private_benchmark_enabled = false;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    if let ChallengeExecutionSpec::PipedStdio(execution) = &mut spec.execution {
        execution.official_session = None;
    }
    create_piped_stdio_bundle(&root, &spec);
    std::fs::write(
        root.join("public/session.json"),
        r#"{"session_name":"public-1","input_files":[],"metadata":null}"#,
    )
    .expect("failed to write empty session inputs");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("empty session input files should fail");
    assert!(error.to_string().contains("non-empty array"));
}

/// Verifies that run manifests cannot declare more than the platform run cap.
#[tokio::test]
async fn run_manifest_rejects_too_many_runs() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-too-many-runs-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    let runs = (0..=crate::challenge_bundle::MAX_CHALLENGE_RUNS_PER_EVALUATION)
        .map(|index| {
            serde_json::json!({
                "run_name": format!("public-{index}"),
                "interface": "stdio",
                "stdin_json": null,
                "stdin_text": "1",
                "input_files": null,
                "output_files": null,
                "metadata": null
            })
        })
        .collect::<Vec<_>>();
    std::fs::write(
        root.join("public/runs.json"),
        serde_json::json!({ "runs": runs }).to_string(),
    )
    .expect("failed to write too-large run manifest");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("too many runs should be rejected");
    assert!(error.to_string().contains("at most 100 runs"));
}

/// Verifies that run names cannot escape evaluator-visible filesystem paths.
#[tokio::test]
async fn run_manifest_rejects_parent_directory_run_name() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-unsafe-run-name-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = false;
    create_bundle(&root, &spec);
    std::fs::write(
        root.join("public/runs.json"),
        r#"{"runs":[{"run_name":"..","interface":"stdio","stdin_json":null,"stdin_text":"1","input_files":null,"output_files":null,"metadata":null}]}"#,
    )
    .expect("failed to write unsafe run manifest");

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    let error = result.expect_err("parent-directory run names should be rejected");
    assert!(error.to_string().contains("run_name"));
}

/// Verifies that enabled private benchmark bundle requires directory.
#[tokio::test]
async fn enabled_private_benchmark_bundle_requires_directory() {
    let root = std::env::temp_dir().join(format!(
        "agentics-bundle-enabled-private-benchmark-{}",
        uuid::Uuid::new_v4()
    ));
    let mut spec = base_spec();
    spec.datasets.private_benchmark_enabled = true;
    spec.datasets.private_benchmark_dir = Some(bundle_path("private-benchmark"));
    create_bundle(&root, &spec);

    let result = validate_challenge_bundle(&root).await;
    drop(std::fs::remove_dir_all(root));

    assert!(result.is_err());
}