@theronap/agnoclast-mcp 0.9.148 → 0.9.149

package/lib/edge_extract.mjs

@@ -289,6 +289,51 @@ export function verbatimIn(needle, haystack) {
   return norm(haystack).includes(n)
 }
 
+// ── SNAPPING A NEAR-QUOTE TO ITS SOURCE (ADR-0059 §4.1, added 2026-09-11) ──────────────────────────
+//
+// 🔴 WHY. The STRAT 421 research brief was lost on two separate runs for one word. The model quoted "…but
+// just have it on your radar. That will be September 22nd as well." The lecture says "on THE radar". The
+// gate above is right to refuse that quote, because a paraphrase is not evidence. But refusing the whole
+// CANDIDATE threw away a real, dated deadline whose source passage is right there, differing by one word.
+//
+// So a quote the gate refuses gets one more chance: find the SOURCE passage it came from, and use the
+// SOURCE's words. The quote must be long (SNAP_MIN_WORDS). Its first and last SNAP_ANCHOR_WORDS words must
+// each appear VERBATIM in the source, in order. The span between them must be within
+// SNAP_LENGTH_TOLERANCE of the quote's length. And exactly ONE such span may exist.
+//
+// ⚠ THIS DOES NOT LOOSEN THE GATE. What is stored is still verbatim source text — the transcript's
+// words, not the model's — so "the evidence sentence is in the source" stays true. What changes is who
+// wrote the evidence. A fabrication does not survive: an invented middle must still sit between two exact,
+// in-order, correctly-spaced anchors, and it is then REPLACED by what the source actually says there.
+// That is the text the reviewer and the person read. Not snapped: short quotes (anchors too weak), a
+// quote splicing distant passages (length check), and an anchor pair that matches in two places.
+const SNAP_ANCHOR_WORDS = 6
+const SNAP_MIN_WORDS = 14
+const SNAP_LENGTH_TOLERANCE = 0.25
+
+/** PURE. The source passage a near-verbatim quote was taken from (whitespace-normalised), or null. */
+export function snapToSource(quote, source) {
+  const q = String(quote ?? '').replace(/\s+/g, ' ').trim()
+  const words = q.split(' ')
+  if (!q || words.length < SNAP_MIN_WORDS) return null
+  const src = String(source ?? '')
+  const esc = (w) => w.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
+  const anchor = (ws) => new RegExp(ws.map(esc).join('\\s+'), 'g')
+  const lo = q.length * (1 - SNAP_LENGTH_TOLERANCE)
+  const hi = q.length * (1 + SNAP_LENGTH_TOLERANCE)
+  const found = new Set()
+  for (const h of src.matchAll(anchor(words.slice(0, SNAP_ANCHOR_WORDS)))) {
+    const tail = anchor(words.slice(-SNAP_ANCHOR_WORDS))
+    tail.lastIndex = h.index + h[0].length
+    for (let t = tail.exec(src); t; t = tail.exec(src)) {
+      const span = src.slice(h.index, t.index + t[0].length).replace(/\s+/g, ' ').trim()
+      if (span.length > hi) break
+      if (span.length >= lo) { found.add(span); break }
+    }
+  }
+  return found.size === 1 ? [...found][0] : null
+}
+
 /**
  * Drop every obligation candidate that cannot be verified against the text the model was shown.
  *
@@ -309,18 +354,24 @@ export function keepVerifiableObligations(items, shown) {
     const evidence = typeof it.evidence === 'string' ? it.evidence.trim() : ''
     const party = typeof it.obligated_party === 'string' ? it.obligated_party.trim() : ''
     if (!subject || !evidence || !OBLIGATION_PARTIES.has(party)) continue
-    if (!verbatimIn(evidence, shown)) continue
+    // Verbatim, or a near-quote snapped to the source's own words (snapToSource). Nothing else.
+    let ev = evidence
+    if (!verbatimIn(ev, shown)) {
+      const snapped = snapToSource(ev, shown)
+      if (!snapped || !verbatimIn(snapped, shown)) continue
+      ev = snapped
+    }
     const due = typeof it.due_at === 'string' && it.due_at.trim() ? it.due_at.trim() : null
     // ADR-0059 §5.2: the WHEN words, which the server resolves into a date in code. Kept only if they
     // appear inside the evidence sentence — the check that stops a paraphrase ("Sept 22" for "the
     // 22nd") or a phrase from some other sentence from deciding the date. Dropped otherwise: the server
     // then falls back to due_at, exactly as it did before phrases existed.
     const phraseRaw = typeof it.due_phrase === 'string' ? it.due_phrase.trim() : ''
-    const due_phrase = phraseRaw && normWs(evidence).includes(normWs(phraseRaw)) ? phraseRaw.slice(0, 120) : null
+    const due_phrase = phraseRaw && normWs(ev).includes(normWs(phraseRaw)) ? phraseRaw.slice(0, 120) : null
     out.push({
       subject: subject.slice(0, 200), due_at: due,
       ...(due_phrase ? { due_phrase } : {}),
-      evidence: evidence.slice(0, 500), obligated_party: party,
+      evidence: ev.slice(0, 500), obligated_party: party,
     })
     if (out.length >= MAX_OBLIGATIONS) break
   }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@theronap/agnoclast-mcp",
-  "version": "0.9.148",
+  "version": "0.9.149",
   "description": "Connect your AI assistant to Cortex — your org's projects, activity, gaps, and directives, scoped to you.",
   "type": "module",
   "bin": {