feat(typesafe): confirmation-loop agent + --typesafe flag (on/off/auto) for A/B

TypeSafe cannot BE an LLM agent — System One does not generate text or call
tools. But it can be the decision brain of a code-owned confirmation loop, and
that is what typesafe_agent.rs is: an ADDITIONAL confirmation strategy.

typesafe_agent.rs — for enumerable classes (XSS, SQLi, open-redirect, path
traversal, SSRF, IDOR): code lists candidate payloads, a TypeSafe Choice picks
the next one given what's been tried, the replay engine sends it for real, a
TypeSafe Noul judges the response, loop until confirmed or exhausted. Edge/WAF
answers are refused. Pure parts (class table, payload templating, id-swap, OAST
substitution, query encoding) are unit-tested; the networked loop is integration.

Wired as a pipeline pass that runs ONLY on findings the LLM path left
unconfirmed or in needs-review (the recall lever) — it can raise a finding to
confirmed with a calibrated probability, never downgrades (the deterministic
layer owns that).

--typesafe on|off|auto (global flag) resolves into the env the pipeline reads,
governing adjudication, CVSS re-grade, agent pruning and this loop together.
`off` runs the identical pipeline without TypeSafe; meta.json records
"typesafe": true|false so a with/without pair is a clean A/B measurement. Web
console gets the same toggle.

381 tests.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

web/public/app.js

@@ -498,6 +498,7 @@ function renderReview() {
     { k: 'Intercept', v: $('#fieldIntercept').value === 'off' ? 'direct' : $('#fieldIntercept').value },
     { k: 'Sandbox', v: $('#fieldSandbox').value ? 'Kali container' : 'host' },
     { k: 'PoC re-validation', v: $('#fieldRevalidatePoc').checked ? 'on' : 'off' },
+    { k: 'TypeSafe', v: $('#fieldTypesafe') ? $('#fieldTypesafe').value : 'auto' },
     { k: 'Compliance', v: (['fieldCompPci', 'fieldCompHipaa', 'fieldCompSoc2'].map((id) => $(`#${id}`).checked && $(`#${id}`).value).filter(Boolean).join(', ')) || 'none' },
     { k: 'Target auth', v: state.auth.header ? 'header set' : (state.auth.roles.length ? `${state.auth.roles.length} role(s)` : 'none') },
   ];
@@ -543,6 +544,7 @@ async function startExploitation() {
     intercept: $('#fieldIntercept').value,
     sandbox: $('#fieldSandbox').value || undefined,
     revalidatePoc: $('#fieldRevalidatePoc').checked,
+    typesafe: $('#fieldTypesafe') ? $('#fieldTypesafe').value : undefined,
     compliance: ['fieldCompPci', 'fieldCompHipaa', 'fieldCompSoc2'].map((id) => $(`#${id}`).checked && $(`#${id}`).value).filter(Boolean),
     tokenLimit: Number($('#fieldTokenLimit').value) || undefined,
     order: $('#fieldOrder').value,

web/public/index.html

@@ -269,6 +269,14 @@
             <div class="field-group"><label class="field-label">PoC re-validation</label>
               <div class="check-row"><input type="checkbox" id="fieldRevalidatePoc" /> <label for="fieldRevalidatePoc">Re-run every PoC; demote what no longer reproduces</label></div>
             </div>
+            <div class="field-group"><label class="field-label" for="fieldTypesafe">TypeSafe (System One)</label>
+              <select class="narrow" id="fieldTypesafe">
+                <option value="auto" selected>auto — on if TYPESAFE_API_KEY set</option>
+                <option value="on">on — calibrated confirmation + adjudication</option>
+                <option value="off">off — same pipeline, no TypeSafe (for A/B)</option>
+              </select>
+              <div class="field-help">Additional confirmation strategy (Choice/Noul over real replays). Off runs the identical pipeline so you can compare with/without.</div>
+            </div>
           </div>
           <div class="field-group">
             <label class="field-label">Compliance mapping</label>