4 Commits
Author SHA1 Message Date
badgerbeesandViet Tran f5917b0e46 fix(backup): harden tenant restore preview and lookup handling (#920)
* fix(backup): harden tenant restore preview and lookup handling

* test(pg): fix hook test migration path

* test(ci): stabilize race coverage tests

* fix(hooks): scope GetByID and allow loopback tests

* fix(backup): harden tenant restore replace/new contracts + race-safe SSRF flag

- Replace mode no longer deletes the tenants row (FK safe vs excluded
  diagnostic tables: traces, activity_logs, usage_snapshots, spans,
  embedding_cache, pairing_requests, paired_devices,
  channel_pending_messages, cron_run_logs). Metadata is preserved in place.
- shouldRestoreTable now excludes tenants for both new and replace modes.
- CLI: add validateTenantRestoreFlags guardrail. mode=new requires
  --new-tenant-slug and rejects --tenant/--tenant-id; upsert/replace warn
  on stray --new-tenant-slug; invalid --mode values rejected. TAB in help
  text fixed; flag descriptions clarified.
- HTTP: resolveRestoreTarget rejects tenant_id for mode=new regardless
  of tenant_slug (matches CLI contract). New i18n key
  MsgRestoreNewModeRejectsTenantID (en/vi/zh).
- security/ssrf: allowLoopbackForTest switched to atomic.Bool so
  concurrent reads from outbound dialers are race-safe.
- Polish: vi backup.json key order matches en/zh; TenantRestoreOptions.Mode
  doc comment documents upsert/replace/new semantics including clone
  behavior for new.
- Tests: unit coverage for validator (12 cases), HTTP guardrails
  (3 cases), shouldRestoreTable replace branch. Integration test
  tests/integration/tenant_restore_replace_test.go regression-guards
  the FK fix using activity_logs seed + DeleteTenantDataForTest helper.

---------

Co-authored-by: Viet Tran <viettranx@gmail.com>
2026-04-16 15:49:50 +07:00
viettranx 8ee540dc7e feat(hooks/builtin): embedded registry + UPSERT seed + store-layer readonly
Phase 04 of hooks Wave 1. Adds infrastructure for shipping canonical hook
rows with the binary; Phase 05 plants the first real builtin (pii-redactor).

- internal/hooks/builtin: loader (//go:embed yaml+js), Seed() with version
  reconciliation (newer embed overwrites DB; downgrade warn-only; preserves
  user's enabled toggle), AllowlistFor() for per-id mutable-field lookup.
  Stable UUIDv5 namespace 082ab084-a25f-52b4-a4a4-eb8a816bd9a8 keys rows
  across restarts so Seed is idempotent on every boot.
- internal/hooks/store.go: ErrBuiltinReadOnly sentinel + WithSeedBypass
  ctx marker (seeder bypasses the guard; users cannot).
- PG + SQLite hook stores: Update/Delete now reject any patch other than
  enabled=* on source='builtin' rows. Fail-closed on GetByID errors.
- Dispatcher: SetBuiltinAllowlistLookup setter lets the gateway wire the
  registry-backed strict allowlist; unwired tests keep Phase 03 permissive
  default. Gateway installs a strip-all lookup first so a Load() failure
  fails closed instead of opening mutation to the permissive default.
- i18n: MsgHookBuiltinReadOnly (en/vi/zh).
- yaml.v3 promoted to direct dep.
- Tests: loader parse + namespace stability + seed idempotency
  (3 boots → N rows), version bump, downgrade detection, operator
  BuiltinDisable escape hatch, PG + SQLite readonly guard round-trip.
2026-04-16 14:17:47 +07:00
viettranx b7592bcacf feat(hooks/script): integrate script handler — FireResult + source-tier gate + migration
Wave 1 Phase 03. Wires the Phase 02 Goja handler into the dispatcher, widens
the DB schema for script + builtin sources, and refactors Dispatcher.Fire to
return a FireResult so builtin-source hooks can mutate event input.

Dispatcher:
- FireResult{Decision, UpdatedToolInput, UpdatedRawInput} replaces the plain
  Decision return. stdDispatcher.runSync keeps a local evMut copy so a
  builtin-source hook's updatedInput flows to downstream hooks in the chain
  and to the caller; non-builtin (source=ui) script mutations are dropped +
  WARN-logged (defense-in-depth; source tier enforced at the dispatcher, not
  only at the handler).
- applyBuiltinMutation + placeholder builtinAllowlistFor (rawInput, toolInput)
  — Phase 04 overrides the allowlist from builtins.yaml.
- noopDispatcher returns FireResult{DecisionAllow}.
- Script mutation tests: TestDispatcher_ScriptMutation_BuiltinSourceApplies,
  TestDispatcher_ScriptMutation_UISourceDenied.

Pipeline + callers (14 Fire sites refactored):
- FireHook wrapper returns FireResult; context_stage.go:52 applies
  UpdatedRawInput → state.Input.Message; tool_stage.go:52 applies
  UpdatedToolInput → tc.Arguments before ExecuteToolCall.
- delegate_bridge + delegate_tool keep the Decision branch; Updated* ignored.
- dispatcher_test / delegate_bridge_test / delegate_tool_hooks_test /
  integration test helpers updated for the new shape.

Validation:
- edition_gate allows HandlerScript on every edition (sandboxed, no shell
  escape surface).
- config.validateHandler rejects empty source, source > 32 KiB, goja compile
  error; validateTimeout rejects on_timeout=ask|defer (reserved).
- Six new config tests cover the script path.

Migrations:
- PG migration 000053 relaxes handler_type + source CHECKs and drops
  uq_hooks_{global,tenant,agent} — scripts routinely want many small hooks
  per event. RequiredSchemaVersion → 53.
- SQLite SchemaVersion → 21; patch 20 is a SELECT 1 placeholder because
  SQLite can't ALTER a CHECK — rebuildAgentHooksV21 runs outside the
  migration tx (parallel to backfillV16) to rename/recreate the table with
  widened CHECKs. schema.sql fresh-DB path matches.
- H9 fix: PGHookStore.Create + SqliteHookStore.Create honor caller-provided
  cfg.ID (uuid.Nil falls back to UUIDv7), unblocking Phase 04 idempotent
  UUIDv5 seed. TestCreateHonorsFixedID on both stores.

Config:
- config.HooksConfig{ScriptConcurrency, ScriptPerTenantConcurrency,
  ScriptCacheSize, BuiltinDisable}; buildHookHandlers wires the script
  handler with the caps from appCfg.Hooks.

Gates green: go build ./... + sqliteonly, go test -race -count=1 across
hooks/pipeline/sqlitestore.
2026-04-16 14:17:47 +07:00
viettranx 3fe5ae0b50 feat(hooks): introduce agent lifecycle hooks foundation with fail-closed blocking semantics
Phase 1 of agent-hooks-system establishes:
- Hook types, matchers, and CEL evaluation engine for event-driven extensibility
- Edition-gated command handlers (Lite disabled) with dedup_key-indexed audit log
- Dual-DB store layer (PostgreSQL + SQLite) with transaction boundary enforcement
- Blocking-event semantics: hook execution failures propagate to agent loop, graceful retry
- Tenant-isolated audit trail with per-hook execution context tracing
- Integration tests verifying store correctness and hook dispatch atomicity

Lays groundwork for post-v3.0 phase-02 (per-tenant onboarding hooks).
2026-04-15 14:25:59 +07:00