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
//! Demonstration of compositional verification through formal verification legos.
//!
//! This example shows how the `elicitation` framework creates a compositionally
//! verified ecosystem where types snap together like legos, with each connection
//! formally proven safe.
//!
//! # The Vision: "Formal Verification Legos"
//!
//! Types implementing `Elicitation` form a verification hierarchy:
//!
//! ```text
//! ┌─────────────────────────────────────────────────┐
//! │ Layer 1: Primitive Types (Manual Kani Proofs) │
//! │ ✓ I8Positive, StringNonEmpty, U8NonZero │
//! └─────────────────┬───────────────────────────────┘
//! │
//! │ implements Elicitation
//! │ #[derive(Elicit)]
//! ↓
//! ┌─────────────────────────────────────────────────┐
//! │ Layer 2: Derived Structs (Compositional Proofs) │
//! │ struct Config { │
//! │ timeout: I8Positive, ← verified │
//! │ retries: U8NonZero, ← verified │
//! │ } │
//! │ ⟹ Config verified by composition ∎ │
//! └─────────────────┬───────────────────────────────┘
//! │
//! │ implements Elicitation
//! │ #[derive(Elicit)]
//! ↓
//! ┌─────────────────────────────────────────────────┐
//! │ Layer 3: Higher-Order Types │
//! │ struct Application { │
//! │ config: Config, ← verified │
//! │ name: StringNonEmpty, ← verified │
//! │ } │
//! │ ⟹ Application verified by composition ∎ │
//! └─────────────────────────────────────────────────┘
//! ```
//!
//! # The "Caged Agent" Property
//!
//! When an LLM is asked to elicit a type `T: Elicitation`:
//! - The type system enforces that `T` is verified
//! - The verification is **non-bypassable** (enforced at compile time)
//! - Invalid states are **unrepresentable** (cannot be constructed)
//!
//! This creates a "cage" where the agent can only produce values that
//! have been mathematically proven to satisfy their contracts.
//!
//! # Running This Example
//!
//! This example doesn't run - it demonstrates compile-time verification:
//!
//! ```bash
//! # Verify compilation (proves compositional structure is valid)
//! cargo check --example compositional_verification
//!
//! # Run Kani proofs (witnesses that verification is sound)
//! cargo kani --harness verify_compositional_legos
//! ```
use Elicit;
use Elicitation;
// ============================================================================
// Layer 1: Primitive Types (Foundation for Composition)
// ============================================================================
// Rust primitives (i32, u16, String, etc.) form the foundation layer.
// The elicitation library provides contract types with manual Kani proofs:
// - I8Positive, U8NonZero: verified in verification/types/kani_proofs/integers.rs
// - StringNonEmpty: verified in verification/types/kani_proofs/strings.rs
//
// This example uses primitives for simplicity, but the compositional
// verification works the same way with contract types.
// ============================================================================
// Layer 2: Derived Structs (Compositional Proofs)
// ============================================================================
use ;
// ============================================================================
// Layer 3: Higher-Order Types (Nested Compositional Proofs)
// ============================================================================
// Example types removed;
// ============================================================================
// Kani Proof Harness: Witness the Compositional Chain
// ============================================================================
// ============================================================================
// Documentation: Understanding the Proof Strategy
// ============================================================================
/// # Tautological Proof by Composition
///
/// The verification strategy is **tautological** (proof by construction):
///
/// ## Base Case (Layer 1)
///
/// Primitive types like `I8Positive` have **manual Kani proofs**:
///
/// ```rust,ignore
/// #[kani::proof]
/// fn verify_i8_positive() {
/// let value: i8 = kani::any();
/// match I8Positive::new(value) {
/// Ok(pos) => assert!(value > 0),
/// Err(_) => assert!(value <= 0),
/// }
/// }
/// ```
///
/// These proofs use symbolic execution to verify **all possible inputs**.
///
/// ## Inductive Case (Layer 2+)
///
/// Derived types inherit verification **by composition**:
///
/// ```rust,ignore
/// #[derive(Elicit)]
/// struct Config {
/// timeout: I8Positive, // ← verified
/// retries: U8NonZero, // ← verified
/// }
///
/// // Generated by #[derive(Elicit)]:
/// impl Elicitation for Config {
/// #[cfg(kani)]
/// fn kani_proof() {
/// I8Positive::kani_proof(); // Verify timeout field
/// U8NonZero::kani_proof(); // Verify retries field
/// // Tautology: all parts verified ⟹ whole verified
/// assert!(true, "Compositional verification");
/// }
/// }
/// ```
///
/// ## The Proof Chain
///
/// 1. **Primitives are verified** (manual Kani proofs with symbolic execution)
/// 2. **Derived types call field proofs** (`#[derive(Elicit)]` generates kani_proof)
/// 3. **Type system enforces** all fields implement `Elicitation`
/// 4. **∴ If compilation succeeds, verification succeeds** ∎
///
/// ## Why This is Non-Bypassable
///
/// The "cage" cannot be escaped because:
/// - **Type system** enforces `Elicitation` trait bounds
/// - **Kani proofs** verify primitives for all possible inputs
/// - **Composition** inherits verification transitively
/// - **Compilation** witnesses the entire chain
///
/// An LLM asked to produce a `Config` cannot:
/// - Bypass type checks (compile-time enforcement)
/// - Produce invalid primitives (Kani proves impossible)
/// - Skip field verification (derive macro enforces)
/// - Construct unverified states (unrepresentable in type system)
///
/// The verification is **proof-carrying**: types carry their own proofs.