feat: zero-touch Agent — central scheduler, central pause, delta sync, discovery-driven installer

Hub: installations.processing_enabled / sync_interval_seconds / last_sync_* (migration 0026).
The first heartbeat of an installation without coverage queues exactly one INITIAL_SYNC;
afterwards one SYNC_INCREMENTAL per interval (idempotent, backs off on failure). One fiscal
job at a time per Agent (FIFO, stale RUNNING recovery). Job completion now updates the
coverage snapshot, blockers and JOB_STARTED/COMPLETED/FAILED events. Panel: Sincronizar
agora / Reprocessar tudo / Pausar / Retomar + status; pause is enforced Hub-side.
Agent: local AGENT_JOBS_ENABLED is ignored (central control); SYNC_INCREMENTAL re-classifies
only new/changed products via a per-product fingerprint+decision cache and rebuilds coverage
from it; results carry operational summary and blockers; Firebird credentials come from a
DPAPI firebird.bin. Setup: read-only SGV Firebird discovery (auto-select only when
unambiguous, wizard chooser otherwise), starts the service, waits for the first heartbeat,
upgrade-in-place mode. All jobs stay read-only (FIREBIRD_WRITES=0). Version 1.1.3.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

agent/src/gti_fiscal_agent/preflight.py

@@ -9,7 +9,7 @@ import os
 import sqlite3
 from pathlib import Path
 
-AGENT_VERSION_DEFAULT = "1.1.2"
+AGENT_VERSION_DEFAULT = "1.1.3"
 REQUIRED_ENV = ("GTI_HUB_URL", "AGENT_TOKEN", "FIREBIRD_DATABASE_PATH", "FB_USER", "FB_PASSWORD")
 REQUIRED_CAPABILITIES = ("supports_ncm", "supports_ipi", "supports_cest")
 FISCAL_MODES = ("DRY_RUN", "AUTO_APPLY")

agent/src/gti_fiscal_agent/state.py

@@ -1,4 +1,5 @@
 """Local SQLite outbox; it has no Firebird dependency or write path."""
+import json
 import sqlite3
 from pathlib import Path
 import uuid
@@ -8,6 +9,9 @@ class AgentState:
         self.connection = sqlite3.connect(Path(path)); self.connection.row_factory = sqlite3.Row
         self.connection.executescript("CREATE TABLE IF NOT EXISTS product_state(product_id TEXT PRIMARY KEY, source_fingerprint TEXT NOT NULL, dataset_version TEXT NOT NULL, engine_version TEXT NOT NULL, status TEXT NOT NULL, last_processed_at TEXT, last_result_hash TEXT, reason_code TEXT); CREATE TABLE IF NOT EXISTS outbox(product_id TEXT PRIMARY KEY, source_fingerprint TEXT NOT NULL, status TEXT NOT NULL, retries INTEGER NOT NULL DEFAULT 0); CREATE TABLE IF NOT EXISTS ipi_write_log(product_id TEXT NOT NULL, source_fingerprint TEXT NOT NULL, decision_status TEXT, ipi_status TEXT, write_status TEXT NOT NULL, before_value TEXT, proposed_value TEXT, result TEXT, reason_code TEXT, timestamp TEXT NOT NULL);")
         self.connection.execute("CREATE TABLE IF NOT EXISTS installation_identity (id INTEGER PRIMARY KEY CHECK(id=1), installation_id TEXT NOT NULL)")
+        # Last Hub decision per product + the source fingerprint it was computed for: what makes
+        # SYNC_INCREMENTAL a real delta (only new/changed products are re-classified).
+        self.connection.execute("CREATE TABLE IF NOT EXISTS decision_cache (product_id TEXT PRIMARY KEY, source_fingerprint TEXT NOT NULL, decision_json TEXT NOT NULL)")
         self.connection.commit()
     def stored_installation_id(self):
         row = self.connection.execute("SELECT installation_id FROM installation_identity WHERE id=1").fetchone()
@@ -35,6 +39,23 @@ class AgentState:
         if row is None or str(uuid.UUID(row["installation_id"])) != value:
             raise ValueError("persisted installation_id conflicts with candidate")
         return value
+    def cached_fingerprint(self, product_id):
+        row = self.connection.execute("SELECT source_fingerprint FROM decision_cache WHERE product_id=?", (product_id,)).fetchone()
+        return row["source_fingerprint"] if row else None
+    def cached_decision(self, product_id):
+        row = self.connection.execute("SELECT decision_json FROM decision_cache WHERE product_id=?", (product_id,)).fetchone()
+        return json.loads(row["decision_json"]) if row else None
+    def remember_decision(self, product_id, source_fingerprint, decision):
+        self.connection.execute("INSERT OR REPLACE INTO decision_cache(product_id,source_fingerprint,decision_json) VALUES(?,?,?)", (product_id, source_fingerprint, json.dumps(decision, default=str)))
+        self.connection.commit()
+    def forget_missing(self, present_product_ids):
+        """Products no longer in the ERP leave the LOCAL cache only; Hub-side history/blockers are never deleted by this."""
+        present = set(present_product_ids)
+        gone = [row["product_id"] for row in self.connection.execute("SELECT product_id FROM decision_cache") if row["product_id"] not in present]
+        for product_id in gone:
+            self.connection.execute("DELETE FROM decision_cache WHERE product_id=?", (product_id,))
+        self.connection.commit()
+        return len(gone)
     def enqueue(self, product_id, source_fingerprint, dataset_version, engine_version):
         row = self.connection.execute("SELECT * FROM product_state WHERE product_id=?", (product_id,)).fetchone()
         if row and tuple(row[key] for key in ("source_fingerprint", "engine_version", "status")) == (source_fingerprint, engine_version, "RESULT_RECEIVED"): return False