# Nuclei query mutation: keryn-nuclei differential profile vs nuclei v3.11.1

> For the same URL, parameter and payloads, does keryn-nuclei's query-differential profile plan the same mutated requests as Nuclei's query replace/single fuzzing?

Source: https://keryn.dev/evals/nuclei-differential/

For the same URL, parameter and payloads, does keryn-nuclei's query-differential profile plan the same mutated requests as Nuclei's query replace/single fuzzing?

| Evaluation detail | Description |
|---|---|
| Keryn engine | keryn-nuclei differential at `e2f8c66973d8`: keryn-nuclei's query-differential profile (crates/keryn-nuclei/src/differential.rs): the planned probe URLs. |
| Reference | nuclei v3.11.1. Oracle only: tools/nuclei/upstream-query-oracle.mjs runs it with -dast against a 127.0.0.1 server that logs request lines; interactsh and update checks off; never shipped. |
| Corpus | keryn-nuclei query-differential lab profiles (crates/keryn-nuclei/tests/fixtures/differential-golden.json): six hand-authored lab cases, boolean payload pairs, restated from `crates/keryn-nuclei/tests/fixtures/differential-golden.json`, `crates/keryn-nuclei/tests/fixtures/upstream-query-oracle.json` |
| Comparison | Per probe: the query parameters this profile plans and the parameters Nuclei sent, both form-decoded ('+' and percent escapes), compared as ordered lists. Identical when equal. Byte encoding is not compared: upstream writes space as '+' and leaves ' and = raw, Keryn writes %20, %27 and %3D. |
| Receipt | [`nuclei-differential.json`](https://keryn.dev/evals/nuclei-differential.json), sha256 `65cc69a3fa68b7a8` |

## Results

| Measure | Value | Count | Denominator |
|---|---|---|---|
| Probes planned exactly as Nuclei sent them (decoded) | **100%** | 24 / 24 | mutated requests across the six lab profiles |

Cases by status: 24 identical.

## What this does and does not measure

- Request planning only: which mutated queries are sent. Verdicts are not compared: Nuclei has no boolean content-differential analyzer; the verdicts are checked against Keryn's own lab apps in crates/keryn-nuclei/tests/differential.rs.
- Keryn also sends a doubled baseline and a closing control per trial; Nuclei sends neither in this mode, so they are not cases here.
- One family (boolean), one query parameter, GET only.
- Restated from the committed fixtures; numbers are not recomputed from a new run.

## All cases (24)

**Show every case**

| Case | Status | Reference rows | Keryn rows | Matching | Missing | Extra |
|---|---|---|---|---|---|---|
| `benign/pair-1-false` | identical |  |  |  |  |  |
| `benign/pair-1-true` | identical |  |  |  |  |  |
| `benign/pair-2-false` | identical |  |  |  |  |  |
| `benign/pair-2-true` | identical |  |  |  |  |  |
| `error-only/pair-1-false` | identical |  |  |  |  |  |
| `error-only/pair-1-true` | identical |  |  |  |  |  |
| `error-only/pair-2-false` | identical |  |  |  |  |  |
| `error-only/pair-2-true` | identical |  |  |  |  |  |
| `fixed/pair-1-false` | identical |  |  |  |  |  |
| `fixed/pair-1-true` | identical |  |  |  |  |  |
| `fixed/pair-2-false` | identical |  |  |  |  |  |
| `fixed/pair-2-true` | identical |  |  |  |  |  |
| `noisy/pair-1-false` | identical |  |  |  |  |  |
| `noisy/pair-1-true` | identical |  |  |  |  |  |
| `noisy/pair-2-false` | identical |  |  |  |  |  |
| `noisy/pair-2-true` | identical |  |  |  |  |  |
| `reflects/pair-1-false` | identical |  |  |  |  |  |
| `reflects/pair-1-true` | identical |  |  |  |  |  |
| `reflects/pair-2-false` | identical |  |  |  |  |  |
| `reflects/pair-2-true` | identical |  |  |  |  |  |
| `vulnerable/pair-1-false` | identical |  |  |  |  |  |
| `vulnerable/pair-1-true` | identical |  |  |  |  |  |
| `vulnerable/pair-2-false` | identical |  |  |  |  |  |
| `vulnerable/pair-2-true` | identical |  |  |  |  |  |
