docs: capture ai quality research posture

docs/receipts/kramer-ai-quality-research-map-20260701.json

@@ -0,0 +1,63 @@
+{
+  "schema": "fable.research_map.receipt.v1",
+  "createdAt": "2026-07-01T22:22:59.132Z",
+  "taskLabel": "kramer-ai-quality-interview",
+  "task": "Map David Kramer/Sentry interview notes on AI coding into Fable engineering constraints",
+  "constraints": [
+    "external transcript/link is advisory until locally verified",
+    "quality/security/liability beat raw code volume",
+    "LLMs help search/summarization/prototyping but do not replace verification",
+    "agents draft, receipts prove, humans approve, guarded 8099/deploy ships",
+    "8098/deploy-webhook remains legacy/internal/stale for fable-agent trust",
+    "do not claim 100x or perfect code generation without receipts",
+    "live receipt > committed receipt > memory summary"
+  ],
+  "existingPrimitives": [
+    "research-map-receipt",
+    "uast-runtime-receipt",
+    "falsification-receipt",
+    "stale-memory-quarantine",
+    "agent-run-trace",
+    "factory-gate"
+  ],
+  "recommendedPlan": [
+    {
+      "name": "Capture grounded AI-coding posture",
+      "goal": "Encode the interview lesson as bounded research: AI can accelerate work, but production code still needs small patches, tests, review, and liability-aware release gates.",
+      "acceptanceCriteria": [
+        "receipt records advisory source status",
+        "receipt rejects 100x/code-volume claims as deploy authority",
+        "receipt preserves no-deploy policy"
+      ],
+      "verification": [
+        "cat docs/receipts/kramer-ai-quality-research-map-20260701.json",
+        "node dist/index.js security scan ."
+      ]
+    },
+    {
+      "name": "Apply lesson to UAST work",
+      "goal": "Keep modifications surgical through receipts/tests and one-language slices instead of a universal writer/runtime.",
+      "acceptanceCriteria": [
+        "no universal syntax guarantee is claimed",
+        "UAST remains receipt-bounded until parser/serializer receipts exist",
+        "patches require native checks before commit"
+      ],
+      "verification": [
+        "npm run -s test -- src/fable5/uast-runtime-receipt.test.ts",
+        "npm run -s build"
+      ]
+    }
+  ],
+  "skipped": [
+    "dashboard",
+    "deploy changes",
+    "vendor tools",
+    "new model architecture",
+    "100x productivity claim",
+    "autonomous production patching",
+    "MCP replacing CLI authority"
+  ],
+  "decision": "ready",
+  "reasons": [],
+  "deployAttempted": false
+}