Repository navigation
Expand file tree
/
Copy pathwitness_e2e.rs
More file actions
288 lines (236 loc) · 10.1 KB
/
Copy pathwitness_e2e.rs
File metadata and controls
288 lines (236 loc) · 10.1 KB
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
//! End-to-end integration tests for ADR-035 capability reports.
//!
//! Tests the full witness → scorecard → governance pipeline with real
//! HMAC-SHA256 signatures, policy enforcement, and deterministic replay.
use rvf_runtime::seed_crypto;
use rvf_runtime::witness::{
GovernancePolicy, ParsedWitness, ScorecardBuilder, WitnessBuilder, WitnessError,
};
use rvf_types::witness::*;
const KEY: &[u8] = b"e2e-test-key-for-witness-bundle!";
fn make_entry(tool: &str, latency: u32, cost: u32, tokens: u32) -> ToolCallEntry {
ToolCallEntry {
action: tool.as_bytes().to_vec(),
args_hash: seed_crypto::seed_content_hash(tool.as_bytes()),
result_hash: [0x00; 8],
latency_ms: latency,
cost_microdollars: cost,
tokens,
policy_check: PolicyCheck::Allowed,
}
}
#[test]
fn full_capability_report_pipeline() {
let mut scorecard = ScorecardBuilder::new();
let policy = GovernancePolicy::autonomous();
// Simulate 5 tasks.
let tasks: Vec<([u8; 16], TaskOutcome, bool)> = vec![
([0x01; 16], TaskOutcome::Solved, true),
([0x02; 16], TaskOutcome::Solved, true),
([0x03; 16], TaskOutcome::Solved, false), // solved but no full evidence
([0x04; 16], TaskOutcome::Failed, true),
([0x05; 16], TaskOutcome::Errored, false),
];
for (task_id, outcome, full_evidence) in &tasks {
let mut builder = WitnessBuilder::new(*task_id, policy.clone())
.with_spec(b"fix issue #123")
.with_outcome(*outcome);
if *full_evidence {
builder = builder
.with_plan(b"1. read\n2. fix\n3. test")
.with_diff(b"--- a/file.rs\n+++ b/file.rs")
.with_test_log(b"test ... ok");
}
builder.record_tool_call(make_entry("Read", 50, 100, 500));
builder.record_tool_call(make_entry("Edit", 100, 200, 1000));
builder.record_tool_call(make_entry("Bash", 3000, 0, 0));
let (payload, header) = builder.build_and_sign(KEY).unwrap();
// Verify signature.
let parsed = ParsedWitness::parse(&payload).unwrap();
parsed.verify_all(KEY, &payload).unwrap();
assert_eq!(header.tool_call_count, 3);
assert_eq!(header.total_cost_microdollars, 300);
scorecard.add_witness(&parsed, 0, 0);
}
let card = scorecard.finish();
assert_eq!(card.total_tasks, 5);
assert_eq!(card.solved, 3);
assert_eq!(card.failed, 1);
assert_eq!(card.errors, 1);
assert!((card.solve_rate - 0.6).abs() < 0.01);
// 2 out of 3 solved have full evidence.
assert!((card.evidence_coverage - 0.6667).abs() < 0.01);
// Total cost = 5 tasks * 300 each = 1500. Solved = 3. 1500/3 = 500.
assert_eq!(card.cost_per_solve_microdollars, 500);
}
#[test]
fn governance_restricted_mode_blocks_writes() {
let policy = GovernancePolicy::restricted();
let mut builder = WitnessBuilder::new([0x10; 16], policy)
.with_spec(b"audit code")
.with_outcome(TaskOutcome::Solved);
// Read is allowed.
let check = builder.record_tool_call(make_entry("Read", 50, 100, 500));
assert_eq!(check, PolicyCheck::Allowed);
// Write is denied.
let check = builder.record_tool_call(make_entry("Write", 100, 200, 1000));
assert_eq!(check, PolicyCheck::Denied);
// Edit is denied.
let check = builder.record_tool_call(make_entry("Edit", 100, 200, 1000));
assert_eq!(check, PolicyCheck::Denied);
// Bash is denied.
let check = builder.record_tool_call(make_entry("Bash", 100, 0, 0));
assert_eq!(check, PolicyCheck::Denied);
assert_eq!(builder.policy_violations.len(), 3);
let (payload, _) = builder.build_and_sign(KEY).unwrap();
let parsed = ParsedWitness::parse(&payload).unwrap();
let entries = parsed.parse_trace();
assert_eq!(entries.len(), 4);
assert_eq!(entries[0].policy_check, PolicyCheck::Allowed);
assert_eq!(entries[1].policy_check, PolicyCheck::Denied);
assert_eq!(entries[2].policy_check, PolicyCheck::Denied);
assert_eq!(entries[3].policy_check, PolicyCheck::Denied);
}
#[test]
fn governance_approved_mode_gates_all() {
let policy = GovernancePolicy::approved();
let mut builder = WitnessBuilder::new([0x20; 16], policy).with_outcome(TaskOutcome::Solved);
let check = builder.record_tool_call(make_entry("Read", 50, 100, 500));
assert_eq!(check, PolicyCheck::Confirmed);
let check = builder.record_tool_call(make_entry("Bash", 100, 0, 0));
assert_eq!(check, PolicyCheck::Confirmed);
// No policy violations — confirmed is not a violation.
assert!(builder.policy_violations.is_empty());
}
#[test]
fn governance_autonomous_with_cost_cap() {
let mut policy = GovernancePolicy::autonomous();
policy.max_cost_microdollars = 500;
let mut builder = WitnessBuilder::new([0x30; 16], policy).with_outcome(TaskOutcome::Solved);
builder.record_tool_call(make_entry("Read", 50, 400, 500));
assert!(builder.policy_violations.is_empty());
builder.record_tool_call(make_entry("Edit", 50, 200, 500));
assert_eq!(builder.policy_violations.len(), 1);
assert!(builder.policy_violations[0].contains("cost budget"));
}
#[test]
fn deterministic_replay_same_bytes() {
let policy = GovernancePolicy::autonomous();
let mut builder = WitnessBuilder::new([0x40; 16], policy)
.with_spec(b"fix bug #42")
.with_plan(b"1. read auth.rs\n2. fix validation")
.with_diff(b"@@ -10,3 +10,5 @@\n+ validate(input);")
.with_test_log(b"test auth::validate ... ok\n3 passed")
.with_outcome(TaskOutcome::Solved);
builder.record_tool_call(make_entry("Read", 50, 100, 500));
builder.record_tool_call(make_entry("Edit", 100, 200, 1000));
builder.record_tool_call(make_entry("Bash", 2000, 0, 0));
let (payload, _) = builder.build_and_sign(KEY).unwrap();
// Parse and extract all sections.
let parsed = ParsedWitness::parse(&payload).unwrap();
parsed.verify_all(KEY, &payload).unwrap();
assert_eq!(parsed.spec.unwrap(), b"fix bug #42");
assert_eq!(parsed.plan.unwrap(), b"1. read auth.rs\n2. fix validation");
assert_eq!(
parsed.diff.unwrap(),
b"@@ -10,3 +10,5 @@\n+ validate(input);"
);
assert_eq!(
parsed.test_log.unwrap(),
b"test auth::validate ... ok\n3 passed"
);
let entries = parsed.parse_trace();
assert_eq!(entries.len(), 3);
assert_eq!(entries[0].action, b"Read");
assert_eq!(entries[1].action, b"Edit");
assert_eq!(entries[2].action, b"Bash");
assert_eq!(entries[0].latency_ms, 50);
assert_eq!(entries[1].cost_microdollars, 200);
assert_eq!(entries[2].tokens, 0);
// The bundle is self-contained evidence.
assert!(parsed.evidence_complete());
}
#[test]
fn tampered_bundle_detected() {
let mut builder = WitnessBuilder::new([0x50; 16], GovernancePolicy::autonomous())
.with_spec(b"original spec")
.with_outcome(TaskOutcome::Solved);
builder.record_tool_call(make_entry("Bash", 100, 0, 0));
let (mut payload, _) = builder.build_and_sign(KEY).unwrap();
// Tamper.
payload[WITNESS_HEADER_SIZE + 10] ^= 0xFF;
let parsed = ParsedWitness::parse(&payload).unwrap();
assert!(parsed.verify_signature(KEY, &payload).is_err());
}
#[test]
fn postmortem_on_failure() {
let builder = WitnessBuilder::new([0x60; 16], GovernancePolicy::autonomous())
.with_spec(b"implement feature X")
.with_diff(b"partial diff")
.with_test_log(b"test feature_x ... FAILED\n0 passed, 1 failed")
.with_postmortem(b"Root cause: missing null check in parser")
.with_outcome(TaskOutcome::Failed);
let (payload, header) = builder.build_and_sign(KEY).unwrap();
assert_eq!(header.outcome, TaskOutcome::Failed as u8);
let parsed = ParsedWitness::parse(&payload).unwrap();
parsed.verify_all(KEY, &payload).unwrap();
assert_eq!(
parsed.postmortem.unwrap(),
b"Root cause: missing null check in parser"
);
}
#[test]
fn scorecard_percentiles() {
let policy = GovernancePolicy::autonomous();
let mut sc = ScorecardBuilder::new();
// Create 20 tasks with varying latencies.
for i in 0..20u8 {
let mut builder = WitnessBuilder::new([i; 16], policy.clone())
.with_spec(b"task")
.with_diff(b"diff")
.with_test_log(b"ok")
.with_outcome(TaskOutcome::Solved);
// Latency: 100, 200, ..., 2000 ms.
builder.record_tool_call(make_entry("Bash", (i as u32 + 1) * 100, 100, 100));
let (payload, _) = builder.build().unwrap();
let parsed = ParsedWitness::parse(&payload).unwrap();
sc.add_witness(&parsed, 0, 0);
}
let card = sc.finish();
assert_eq!(card.total_tasks, 20);
assert_eq!(card.solved, 20);
assert!((card.solve_rate - 1.0).abs() < 0.01);
assert!((card.evidence_coverage - 1.0).abs() < 0.01);
// Median of 100..2000 (step 100) = ~1000
assert!(card.median_latency_ms >= 900 && card.median_latency_ms <= 1100);
// p95 should be near 1900-2000.
assert!(card.p95_latency_ms >= 1800);
}
#[test]
fn rollback_tracking() {
let mut builder = WitnessBuilder::new([0x70; 16], GovernancePolicy::autonomous())
.with_outcome(TaskOutcome::Solved);
builder.record_rollback();
builder.record_rollback();
assert_eq!(builder.rollback_count, 2);
let (payload, _) = builder.build().unwrap();
let parsed = ParsedWitness::parse(&payload).unwrap();
let mut sc = ScorecardBuilder::new();
sc.add_witness(&parsed, 0, 2);
let card = sc.finish();
assert_eq!(card.rollback_count, 2);
}
#[test]
fn zero_policy_violations_in_autonomous() {
let policy = GovernancePolicy::autonomous();
let mut total_violations = 0u32;
for i in 0..100u8 {
let mut builder =
WitnessBuilder::new([i; 16], policy.clone()).with_outcome(TaskOutcome::Solved);
builder.record_tool_call(make_entry("Read", 10, 10, 10));
builder.record_tool_call(make_entry("Edit", 10, 10, 10));
builder.record_tool_call(make_entry("Bash", 10, 10, 10));
total_violations += builder.policy_violations.len() as u32;
}
assert_eq!(total_violations, 0, "zero policy violations in 100 runs");
}