@futdevpro/nts-dynamo 1.15.64 → 1.15.68

package/__documentations/2026-06-01-fr047-p2p3-function-calling.md

@@ -1,81 +1,81 @@
-# FR-047 P2+P3 — Function-calling agent-loop + OpenAI-wire translator (2026-06-01)
-
-**Commits**: `de6f79e` — v01.15.36 → v01.15.37 (agent-loop + OpenAI translator)
-            `514153f` — → v01.15.40 (critical reasoning-model fix; see below)
-**Parent**: `__agent/feature-requests/FR-047-phase-2-agent-loop-service-base.md` + `FR-047-phase-3-openai-fdpai-local.md`
-**Pattern**: `documentations/guidelines/development/ai-function-calling.md`
-**Depends on**: `@futdevpro/fsm-dynamo` ≥ 1.15.17 (tool vocabulary + registry; bumped to 1.15.18 here)
-
-## What
-
-Implements the provider-agnostic function-calling agent-loop in `DyNTS_AI_LLM_ServiceBase` and the
-first concrete provider translator in `DyNTS_OAI_LLM_ServiceBase`. After this, OpenAI tool calling
-works end-to-end through the strategic agnostic layer.
-
-## Why
-
-FR-047 P1 (fsm) shipped the vocabulary; the execution layer had the explicit TODO comment in
-`getMessageCreateInput` ("'llm-tool-use' categoria a tools bevezetésekor — jövőbeli FR"). This is that FR.
-The proven loop logic is ported from the legacy `FDPNTS_GPT_ControlService.getAnswerWithTools`.
-
-## Design
-
-### Agent-loop in `DyNTS_AI_LLM_ServiceBase` (written ONCE, concrete)
-- `requestWithTools({ conversation, tools, toolHandlers, settings?, issuer, maxIterations? })` — public entry.
-  Tools are a **request parameter** (not on settings). Resolves model → capability pre-check → `runToolLoop`.
-- `runToolLoop` — the loop: `callModelWithTools` → if `toolCalls`, push assistant-tool-call message,
-  run each handler (`runToolCall`), push `role:'tool'` result messages, repeat; exit on no-tool-call
-  final answer or `maxIterations` (default 8 → `DyFM_Error` `DyNTS-AILSB-TL0`).
-- `runToolCall` — **never throws**; missing/throwing handler → `ERROR …` string with `isError:true` (no `[object Object]`).
-- `assertToolsSupported(modelId)` — uses `DyFM_AI_ModelRegistry_Util.modelSupportsTools(getModelRegistry(), id)`;
-  rejects BEFORE any API call if unsupported (critical for Local).
-- `callModelWithTools` + `getModelRegistry` — **non-abstract safe defaults** (throw / `[]`); providers override.
-  Chosen over `abstract` so NO existing concrete subclass breaks (incremental rollout).
-
-### OpenAI translator in `DyNTS_OAI_LLM_ServiceBase`
-- `getModelRegistry()` → `DyFM_OAI_Models`.
-- `callModelWithTools` — builds the native request (system + `toOpenAIMessage`-mapped conversation incl.
-  tool-call/tool-result messages + `toOpenAITool`), calls the SDK, normalizes the response (`toolCalls`
-  from `message.tool_calls`, args JSON-parsed), emits `'llm-tool-use'` cost-event.
-- `toOpenAIMessage` — maps `role:'tool'` → `{role:'tool',tool_call_id,content}`, assistant-with-toolCalls →
-  `{role:'assistant',content|null,tool_calls:[…]}`.
-
-### Refinement vs the FR plan
-The registry lookup util takes the **registry array as a param** (`(models, id)`), not `(provider, id)` —
-so the fsm core never imports a provider submodule. The provider service supplies its own registry.
-
-## CRITICAL fix (`514153f`, v01.15.40) — found in code review
-
-The registry's tool-capable models (gpt-5.5, gpt-5.4-mini, o3-pro) are **reasoning models** that reject
-`temperature`/`top_p`/`frequency_penalty`/`presence_penalty` with HTTP 400. `callModelWithTools` sent them
-unconditionally (default temperature=0) → **every tool call 400'd on the flagship models**. Fix:
-- `isReasoningModel(modelId)` = `/^gpt-5/ || /^o[0-9]/`; reasoning → omit the 4 sampling params.
-- Omit `tools`/`tool_choice` when the tool array is empty (empty `[]` → 400).
-- Clear error when no model is configured.
-This is the local OpenAI-specific application of the FR-048 per-model-settings principle.
-See `[[reference_oai_reasoning_models_reject_sampling_params]]`.
-
-## Spec coverage
-
-`oai-llm.service-base.tools.spec.ts` (mocked OpenAI client, no network) — 6 specs:
-loop converges / never-throw on handler error / capability-reject for embedding model /
-reasoning omits sampling params / non-reasoning sends them / empty-tools omitted.
-`npx jasmine`: **1256/0** (146 pending pre-existing).
-
-## What's NOT in this commit
-
-- Anthropic / Google / FdpAI / Local LLM bases + translators (FR-047 P4/P5).
-- The NON-tool path (`getMessageCreateInput`/`resolveMessage`) still forwards sampling params → reasoning
-  completions would 400 (pre-existing; FR-048 / separate fix).
-- Adventor migration (FR-047 P6).
-- FR-048 general per-model settings reconciler.
-
-## Consumer pattern
-
-See `documentations/guidelines/development/ai-function-calling.md` ("How to USE it").
-
-## Cross-references
-
-- `documentations/guidelines/development/ai-function-calling.md` (pattern + provider-translator how-to)
-- `NPM-packages/dynamo-fsm/__documentations/2026-06-01-fr047-p1-function-calling-vocabulary.md` (vocabulary)
-- legacy `NPM-packages/fdp-templates-nts/src/_modules/gpt/_services/gpt-handler.control-service.ts` (logic source)
+# FR-047 P2+P3 — Function-calling agent-loop + OpenAI-wire translator (2026-06-01)
+
+**Commits**: `de6f79e` — v01.15.36 → v01.15.37 (agent-loop + OpenAI translator)
+            `514153f` — → v01.15.40 (critical reasoning-model fix; see below)
+**Parent**: `__agent/feature-requests/FR-047-phase-2-agent-loop-service-base.md` + `FR-047-phase-3-openai-fdpai-local.md`
+**Pattern**: `documentations/guidelines/development/ai-function-calling.md`
+**Depends on**: `@futdevpro/fsm-dynamo` ≥ 1.15.17 (tool vocabulary + registry; bumped to 1.15.18 here)
+
+## What
+
+Implements the provider-agnostic function-calling agent-loop in `DyNTS_AI_LLM_ServiceBase` and the
+first concrete provider translator in `DyNTS_OAI_LLM_ServiceBase`. After this, OpenAI tool calling
+works end-to-end through the strategic agnostic layer.
+
+## Why
+
+FR-047 P1 (fsm) shipped the vocabulary; the execution layer had the explicit TODO comment in
+`getMessageCreateInput` ("'llm-tool-use' categoria a tools bevezetésekor — jövőbeli FR"). This is that FR.
+The proven loop logic is ported from the legacy `FDPNTS_GPT_ControlService.getAnswerWithTools`.
+
+## Design
+
+### Agent-loop in `DyNTS_AI_LLM_ServiceBase` (written ONCE, concrete)
+- `requestWithTools({ conversation, tools, toolHandlers, settings?, issuer, maxIterations? })` — public entry.
+  Tools are a **request parameter** (not on settings). Resolves model → capability pre-check → `runToolLoop`.
+- `runToolLoop` — the loop: `callModelWithTools` → if `toolCalls`, push assistant-tool-call message,
+  run each handler (`runToolCall`), push `role:'tool'` result messages, repeat; exit on no-tool-call
+  final answer or `maxIterations` (default 8 → `DyFM_Error` `DyNTS-AILSB-TL0`).
+- `runToolCall` — **never throws**; missing/throwing handler → `ERROR …` string with `isError:true` (no `[object Object]`).
+- `assertToolsSupported(modelId)` — uses `DyFM_AI_ModelRegistry_Util.modelSupportsTools(getModelRegistry(), id)`;
+  rejects BEFORE any API call if unsupported (critical for Local).
+- `callModelWithTools` + `getModelRegistry` — **non-abstract safe defaults** (throw / `[]`); providers override.
+  Chosen over `abstract` so NO existing concrete subclass breaks (incremental rollout).
+
+### OpenAI translator in `DyNTS_OAI_LLM_ServiceBase`
+- `getModelRegistry()` → `DyFM_OAI_Models`.
+- `callModelWithTools` — builds the native request (system + `toOpenAIMessage`-mapped conversation incl.
+  tool-call/tool-result messages + `toOpenAITool`), calls the SDK, normalizes the response (`toolCalls`
+  from `message.tool_calls`, args JSON-parsed), emits `'llm-tool-use'` cost-event.
+- `toOpenAIMessage` — maps `role:'tool'` → `{role:'tool',tool_call_id,content}`, assistant-with-toolCalls →
+  `{role:'assistant',content|null,tool_calls:[…]}`.
+
+### Refinement vs the FR plan
+The registry lookup util takes the **registry array as a param** (`(models, id)`), not `(provider, id)` —
+so the fsm core never imports a provider submodule. The provider service supplies its own registry.
+
+## CRITICAL fix (`514153f`, v01.15.40) — found in code review
+
+The registry's tool-capable models (gpt-5.5, gpt-5.4-mini, o3-pro) are **reasoning models** that reject
+`temperature`/`top_p`/`frequency_penalty`/`presence_penalty` with HTTP 400. `callModelWithTools` sent them
+unconditionally (default temperature=0) → **every tool call 400'd on the flagship models**. Fix:
+- `isReasoningModel(modelId)` = `/^gpt-5/ || /^o[0-9]/`; reasoning → omit the 4 sampling params.
+- Omit `tools`/`tool_choice` when the tool array is empty (empty `[]` → 400).
+- Clear error when no model is configured.
+This is the local OpenAI-specific application of the FR-048 per-model-settings principle.
+See `[[reference_oai_reasoning_models_reject_sampling_params]]`.
+
+## Spec coverage
+
+`oai-llm.service-base.tools.spec.ts` (mocked OpenAI client, no network) — 6 specs:
+loop converges / never-throw on handler error / capability-reject for embedding model /
+reasoning omits sampling params / non-reasoning sends them / empty-tools omitted.
+`npx jasmine`: **1256/0** (146 pending pre-existing).
+
+## What's NOT in this commit
+
+- Anthropic / Google / FdpAI / Local LLM bases + translators (FR-047 P4/P5).
+- The NON-tool path (`getMessageCreateInput`/`resolveMessage`) still forwards sampling params → reasoning
+  completions would 400 (pre-existing; FR-048 / separate fix).
+- Adventor migration (FR-047 P6).
+- FR-048 general per-model settings reconciler.
+
+## Consumer pattern
+
+See `documentations/guidelines/development/ai-function-calling.md` ("How to USE it").
+
+## Cross-references
+
+- `documentations/guidelines/development/ai-function-calling.md` (pattern + provider-translator how-to)
+- `NPM-packages/dynamo-fsm/__documentations/2026-06-01-fr047-p1-function-calling-vocabulary.md` (vocabulary)
+- legacy `NPM-packages/fdp-templates-nts/src/_modules/gpt/_services/gpt-handler.control-service.ts` (logic source)

package/__documentations/2026-06-15-fr193-memory-guard.md

@@ -1,73 +1,73 @@
-# FR-193 — `DyNTS_MemoryGuard` (bedrock OOM korai-figyelmeztetés)
-
-**Csomag:** `@futdevpro/nts-dynamo` · **Dátum:** 2026-06-15
-
-## Probléma
-
-A Node heap-OOM (`--max-old-space-size` túllépés — pl. egy óriási `JSON.parse`, mint az
-Overseernél a CCAP több-tíz-MB-os `cdpReport`-ja) a process-t **azonnal megöli**. A
-`process.on('uncaughtException')` NEM kapja el — a crash pillanatában már **nincs mód a
-hibát rögzíteni**. Eredmény: a szerver „csak eltűnik", nyom nélkül, és a kézi diagnózis
-nehéz. Az EGYETLEN megbízható idő a rögzítésre a crash **ELŐTT** van.
-
-## Megoldás — `DyNTS_MemoryGuard`
-
-Bedrock, self-installing heap-watchdog **minden nts-szerverhez**. Periodikusan figyeli a
-heap-kihasználtságot, és a `--max-old-space-size` plafonhoz közeledve **tartós rekordot**
-hagy — mielőtt a fatal OOM bekövetkezne.
-
-- **Megfigyel:** `v8.getHeapStatistics().heap_size_limit` (a valódi plafon) vs
-  `process.memoryUsage().heapUsed`, default 10s-enként.
-- **Rögzít:** küszöb-átlépéskor `DyFM_Log` + `DyNTS_GlobalService.globalErrorHandler`
-  (az Errors-rendszer + Discord, ha bekötve). `warning` → `DyFM_ErrorLevel.warning`,
-  `critical` → `.critical`. Plusz egy in-memory ring-buffer (`getHistory()`) +
-  `getStatus()` (állapot + csúcsok) egy esetleges status-endpointhoz.
-- **Hiszterézis:** eseményt CSAK állapot-váltáskor jelez (normal→warning→critical), és a
-  `recoveryMargin`-nyit a warning alá esve jelez `recovered`-et → nincs küszöb-körüli spam.
-- **NEM csinál:** nem állítja le/újraindítja a process-t (az a konténer `restart:` policy
-  dolga), és nem próbálja elkapni a fatal OOM-ot (nem lehet). Tisztán megfigyelés +
-  rögzítés + opcionális `onCritical` hook (amivel a fogyasztó terhet dobhat).
-- **Biztonság:** minden út `try/catch`-elt — egy figyelő-réteg SOHA nem dönthet be egy
-  szervert. A timer `unref()`-elt (nem tartja életben a process-t).
-
-## Használat
-
-**Automatikus (default-on):** a base `App` startup feltelepíti, ha
-`DyNTS_global_settings.memoryGuard.enabled === true` (alapértelmezés szerint igen). Tehát
-**minden nts-szerver megkapja** az nts-dynamo bump után — kód nélkül.
-
-**Explicit / testreszabott:**
-```ts
-import { DyNTS_MemoryGuard } from '@futdevpro/nts-dynamo';
-
-DyNTS_MemoryGuard.getInstance().install({
-  heapWarningThreshold: 80,
-  heapCriticalThreshold: 92,
-  onCritical: (e) => { /* pl. terhelés-dobás / külső riasztás */ },
-});
-```
-
-**Konfiguráció** (`DyNTS_global_settings.memoryGuard`, defaultok):
-| mező | default | jelentés |
-|------|---------|----------|
-| `enabled` | `true` | a base App feltelepíti |
-| `pollIntervalMs` | `10000` | poll-gyakoriság |
-| `heapWarningThreshold` | `85` | warning a heap-plafon %-ában |
-| `heapCriticalThreshold` | `95` | critical % |
-| `recoveryMargin` | `10` | a warning alatti margó a `recovered`-hez |
-| `maxHistoryCount` | `100` | megőrzött események |
-
-## Az OOM-self-healing nagyobb képe
-
-Ez a **„rögzítés" rétege** (record). A **„restart" rétege** a konténer-szintű
-`restart: on-failure:N` + `mem_limit` (compose, fdp-devops) — együtt adják a teljes
-self-healinget: a guard nyomot hagy a near-OOM-ról, a restart-policy auto-helyreállít a
-crash után. A **forrás-prevenció** (a túl nagy payload parse-olásának elkerülése) külön
-réteg (runner-oldali report-cap / body-limit).
-
-## Fájlok
-- `src/_services/core/memory-guard.service.ts` + `.spec.ts` (11 spec)
-- `src/_models/interfaces/global-settings.interface.ts` (`memoryGuard` mező)
-- `src/_collections/global-settings.const.ts` (defaultok, enabled:true)
-- `src/_services/server/app.server.ts` (guarded auto-install a startup-ban)
-- `src/index.ts` (fő-export)
+# FR-193 — `DyNTS_MemoryGuard` (bedrock OOM korai-figyelmeztetés)
+
+**Csomag:** `@futdevpro/nts-dynamo` · **Dátum:** 2026-06-15
+
+## Probléma
+
+A Node heap-OOM (`--max-old-space-size` túllépés — pl. egy óriási `JSON.parse`, mint az
+Overseernél a CCAP több-tíz-MB-os `cdpReport`-ja) a process-t **azonnal megöli**. A
+`process.on('uncaughtException')` NEM kapja el — a crash pillanatában már **nincs mód a
+hibát rögzíteni**. Eredmény: a szerver „csak eltűnik", nyom nélkül, és a kézi diagnózis
+nehéz. Az EGYETLEN megbízható idő a rögzítésre a crash **ELŐTT** van.
+
+## Megoldás — `DyNTS_MemoryGuard`
+
+Bedrock, self-installing heap-watchdog **minden nts-szerverhez**. Periodikusan figyeli a
+heap-kihasználtságot, és a `--max-old-space-size` plafonhoz közeledve **tartós rekordot**
+hagy — mielőtt a fatal OOM bekövetkezne.
+
+- **Megfigyel:** `v8.getHeapStatistics().heap_size_limit` (a valódi plafon) vs
+  `process.memoryUsage().heapUsed`, default 10s-enként.
+- **Rögzít:** küszöb-átlépéskor `DyFM_Log` + `DyNTS_GlobalService.globalErrorHandler`
+  (az Errors-rendszer + Discord, ha bekötve). `warning` → `DyFM_ErrorLevel.warning`,
+  `critical` → `.critical`. Plusz egy in-memory ring-buffer (`getHistory()`) +
+  `getStatus()` (állapot + csúcsok) egy esetleges status-endpointhoz.
+- **Hiszterézis:** eseményt CSAK állapot-váltáskor jelez (normal→warning→critical), és a
+  `recoveryMargin`-nyit a warning alá esve jelez `recovered`-et → nincs küszöb-körüli spam.
+- **NEM csinál:** nem állítja le/újraindítja a process-t (az a konténer `restart:` policy
+  dolga), és nem próbálja elkapni a fatal OOM-ot (nem lehet). Tisztán megfigyelés +
+  rögzítés + opcionális `onCritical` hook (amivel a fogyasztó terhet dobhat).
+- **Biztonság:** minden út `try/catch`-elt — egy figyelő-réteg SOHA nem dönthet be egy
+  szervert. A timer `unref()`-elt (nem tartja életben a process-t).
+
+## Használat
+
+**Automatikus (default-on):** a base `App` startup feltelepíti, ha
+`DyNTS_global_settings.memoryGuard.enabled === true` (alapértelmezés szerint igen). Tehát
+**minden nts-szerver megkapja** az nts-dynamo bump után — kód nélkül.
+
+**Explicit / testreszabott:**
+```ts
+import { DyNTS_MemoryGuard } from '@futdevpro/nts-dynamo';
+
+DyNTS_MemoryGuard.getInstance().install({
+  heapWarningThreshold: 80,
+  heapCriticalThreshold: 92,
+  onCritical: (e) => { /* pl. terhelés-dobás / külső riasztás */ },
+});
+```
+
+**Konfiguráció** (`DyNTS_global_settings.memoryGuard`, defaultok):
+| mező | default | jelentés |
+|------|---------|----------|
+| `enabled` | `true` | a base App feltelepíti |
+| `pollIntervalMs` | `10000` | poll-gyakoriság |
+| `heapWarningThreshold` | `85` | warning a heap-plafon %-ában |
+| `heapCriticalThreshold` | `95` | critical % |
+| `recoveryMargin` | `10` | a warning alatti margó a `recovered`-hez |
+| `maxHistoryCount` | `100` | megőrzött események |
+
+## Az OOM-self-healing nagyobb képe
+
+Ez a **„rögzítés" rétege** (record). A **„restart" rétege** a konténer-szintű
+`restart: on-failure:N` + `mem_limit` (compose, fdp-devops) — együtt adják a teljes
+self-healinget: a guard nyomot hagy a near-OOM-ról, a restart-policy auto-helyreállít a
+crash után. A **forrás-prevenció** (a túl nagy payload parse-olásának elkerülése) külön
+réteg (runner-oldali report-cap / body-limit).
+
+## Fájlok
+- `src/_services/core/memory-guard.service.ts` + `.spec.ts` (11 spec)
+- `src/_models/interfaces/global-settings.interface.ts` (`memoryGuard` mező)
+- `src/_collections/global-settings.const.ts` (defaultok, enabled:true)
+- `src/_services/server/app.server.ts` (guarded auto-install a startup-ban)
+- `src/index.ts` (fő-export)

package/__documentations/nts-integration-tests-2026-03-17.md

@@ -1,26 +1,26 @@
-# NTS integration tests (unified host)
-
-**Date:** 2026-03-17
-
-## Summary
-
-First integration tests for dynamo-nts were added to cover the unified host behaviour (static client, API base path, socket path, default 404 page).
-
-## Additions
-
-- **Integration mock:** [src/_modules/mock/app-integration-test.mock.ts](../src/_modules/mock/app-integration-test.mock.ts)  
-  `DyNTS_AppIntegrationTest_Mock` extends `DyNTS_AppWbMock_Mock`, runs without DB, and provides:
-  - `getAppParams()` with no `dbName`/`dbUri`
-  - `getPortSettings()` with `httpPort: 54536`, `socketPath: '/socket'`
-  - `getApiBasePath()` returning `'/api'`
-  - `getStaticClientSettings()` returning `root` from static property `integrationStaticRoot` (set by the spec), no `fallbackPath` (so 404 uses default HTML)
-
-- **Integration spec:** [src/_modules/socket/app-extended.integration.spec.ts](../src/_modules/socket/app-extended.integration.spec.ts)  
-  - Uses a temp dir for static files and starts the app in `beforeAll`; stops app and removes temp dir in `afterAll`.
-  - Timeout set to 15s for this suite.
-  - Tests: app starts (`started === true`), GET non-existent path → 404 + default “Page not found” HTML, GET `/index.html` → 200 + static content, GET `/api/test-0/test-base` and `/api/test-0/test-simple` → 200 with expected body.
-
-## Conventions
-
-- Same test style as existing NTS specs: Jasmine, `*.spec.ts`, run via `pnpm test` (build + jasmine).
-- Port **54536** must be free when the integration spec runs (used by the mock app).
+# NTS integration tests (unified host)
+
+**Date:** 2026-03-17
+
+## Summary
+
+First integration tests for dynamo-nts were added to cover the unified host behaviour (static client, API base path, socket path, default 404 page).
+
+## Additions
+
+- **Integration mock:** [src/_modules/mock/app-integration-test.mock.ts](../src/_modules/mock/app-integration-test.mock.ts)  
+  `DyNTS_AppIntegrationTest_Mock` extends `DyNTS_AppWbMock_Mock`, runs without DB, and provides:
+  - `getAppParams()` with no `dbName`/`dbUri`
+  - `getPortSettings()` with `httpPort: 54536`, `socketPath: '/socket'`
+  - `getApiBasePath()` returning `'/api'`
+  - `getStaticClientSettings()` returning `root` from static property `integrationStaticRoot` (set by the spec), no `fallbackPath` (so 404 uses default HTML)
+
+- **Integration spec:** [src/_modules/socket/app-extended.integration.spec.ts](../src/_modules/socket/app-extended.integration.spec.ts)  
+  - Uses a temp dir for static files and starts the app in `beforeAll`; stops app and removes temp dir in `afterAll`.
+  - Timeout set to 15s for this suite.
+  - Tests: app starts (`started === true`), GET non-existent path → 404 + default “Page not found” HTML, GET `/index.html` → 200 + static content, GET `/api/test-0/test-base` and `/api/test-0/test-simple` → 200 with expected body.
+
+## Conventions
+
+- Same test style as existing NTS specs: Jasmine, `*.spec.ts`, run via `pnpm test` (build + jasmine).
+- Port **54536** must be free when the integration spec runs (used by the mock app).

package/_specifications/BACKLOG.md

@@ -1,92 +1,92 @@
-# BACKLOG.md
-
-> STRICT PATTERN (MUST FOLLOW EXACTLY)
-> - Entry header: "- [TYPE] (BL-YYYYMMDD-###) TITLE"
-> - Immediately followed by (exact keys, colon-space, indented 2 spaces):
->   - "  status: <EMOJI status-key>"
->   - "  priority: <low|medium|high|urgent>"
->   - "  source: <user|system|assistant>"
->   - "  area: <ui|backend|infra|docs|tests|ux|general>"
->   - "  details: <single-line summary>"
-> - NO blank line between header and fields (parser reads i+1..i+5).
-> - One blank line between entries (readability).
-> - TYPE values: FEATURE, BUG, IMPROVEMENT, QUESTION, RESEARCH, TASK
-> - STATUS EMOJI: ❌ ⏳ 🔄 ✅ ⚠️ ❓
-> - IDs: unique, format BL-YYYYMMDD-### (increment ###).
-
-- [IMPROVEMENT] (BL-20260518-003) LVS hybrid: BM25 corpus + numberOfCandidates pre-filter cache (FR-004 followup)
-  status: ⏳ pending
-  priority: low
-  source: assistant
-  area: backend
-  details: A jelenlegi hybrid search minden vectorSearch hívásnál ujraepiti a BM25 corpus-t (O(N×|doc|)) és a cosine-t ALL candidate-re futtatja, nem csak top-N-re. ~100..10000 doc LVS-pool-ra elfogadható (ms-mp), de 100K+ doc + ~1KB/doc → ~100MB token allocation per query + GC pressure. Two-pronged optim: (1) opt-in BM25 index cache (corpus-key alapján, invalidate on data write), (2) `numberOfCandidates` érdemi haszna — cosine top-numberOfCandidates pre-filter, BM25 csak ezeken fut. CCAP RAG batch-hívásokra különösen releváns. Spec: lvs-bm25.util.ts (cache layer) + lvs-local-vector-search.data-service.ts (pre-filter branch hybrid módban).
-
-- [IMPROVEMENT] (BL-20260518-002) LVS hybrid: Partial<hybridWeight> API ergonomics (FR-004 followup)
-  status: ⏳ pending
-  priority: low
-  source: assistant
-  area: backend
-  details: A jelenlegi `vectorSearch({ ..., hybridWeight: { vector: number; text: number } })` típus mindkét mezőt kötelezővé teszi ha `hybridWeight` megadva. Ergonomikusabb: `hybridWeight?: Partial<{ vector: number; text: number }>` — `{ vector: 0.8 }` esetén `text` default 0.5 (vagy `1 - vector` semantic-szal). Default-merge a `vectorSearch`-ben: `wVector = hybridWeight?.vector ?? 0.5`, `wText = hybridWeight?.text ?? (1 - wVector)`. Non-breaking (jelenlegi `{vector:0.7,text:0.3}` továbbra is valid). 1-2 új spec a partial override-ra. Fájl: lvs-local-vector-search.data-service.ts:130 + spec extension.
-
-- [BUG] (BL-20260518-001) FR-002 cost-event batch test mock/assertion mismatch
-  status: ✅ done
-  priority: medium
-  source: assistant
-  area: tests
-  details: Fix opció (a) választva — mockEmbeddingsCreate.and.returnValue per-test override 2-elem data + usage:{prompt_tokens:10,total_tokens:10} a batch szándékát tükrözve. A spec-fájl `it('| createEmbeddings emits 1 cost-event with embedding-batch callType')` (oai-embedding.control-service.spec.ts:~275-290) override-ot kapott a top-level beforeEach (line 22-47) `prompt_tokens:5` default fölé. Verifikáció: `npx tsc` (build regenerate) + `npx jasmine` → 1232 specs / 0 failures. **Process root-cause azonosítva**: `spec/support/jasmine.json` `spec_dir: build/` — a jasmine a TS-fordított JS-ből fut, NEM a src-ből. A 08defee pre-commit smoke-ban `npx tsc --noEmit` (csak type-check, NEM emit) + `npx jasmine` futott → a `build/` STALE volt (az új teszt-em NEM volt benne), így a régi 1195 spec-pool fut zöldön, az új 10 spec-em sem assertelt. **5-stage cycle Test stage policy** (saját magamra): pre-commit jasmine futtatás MINDIG `npx tsc` (emit) ELŐZZE MEG, NEM `--noEmit`-tel ellenőrizni. Followup ez a process-tanulság a memóriába (feedback_test_locally_first kiegészítés).
-
-- [FEATURE] (BL-20260517-002) Per-call cost-event hook on AI services (FR-002)
-  status: ✅ done
-  priority: high
-  source: user
-  area: backend
-  details: CCAP CT_ cost-tracking blokkoló (MP3-C, MP4-D, MP6-D). Non-breaking, additív: új `DyNTS_AI_CostEvent` interface (callType, provider, model, tokensUsed: {input, output?, total}, durationMs, issuer, timestamp, estimatedCostUsd?) + `DyNTS_AI_CostEventCallback` type alias. `DyNTS_AI_Provider_ServiceBase`-en protected `onCostEvent?` field + protected `emitCostEvent(event)` safe helper (try/catch, log warn — callback hibája NEM akasztja meg az AI-call-t). OAI emit-helyek: `DyNTS_OAI_Embedding_ControlService.createEmbedding` ('embedding-single'), `.createEmbeddings` ('embedding-batch'), `DyNTS_OAI_LLM_ServiceBase.resolveMessage` ('llm-completion' ha 1 user-msg, 'llm-chat' ha több), `DyNTS_OAI_LLM_Chat_ServiceBase.resolveConversation` ('llm-chat'). Timing: Date.now() start/end, durationMs >= 0. Provider/model auto-resolved (this.aiProvider + String(callInput.model)). Constructor opt: `DyFM_OAI_Settings & { onCostEvent? }` intersection. `estimatedCostUsd` Dynamo-szinten mindig undefined (nincs pricing registry); CCAP CT_ consumer tölti ki. 'llm-tool-use' callType jövőbeli FR. Anthropic / FDP-AI provider-ek csak az alap-osztály onCostEvent mezőt öröklik (még nem implementálnak embedding/LLM-et, saját emit-helyek külön FR). oai-vector-data.service.ts SKIPPED — fájl nem létezik (a spec hivatkozása konceptuális). 10 új jasmine teszt (4 base + 4 embedding + 2 LLM + 1 constructor). Full suite: 1195/0. Spec: __agent/feature-requests/FR-002-per-call-cost-event-hook.md (workspace).
-
-- [FEATURE] (BL-20260517-003) LVS hybrid search: cosine + BM25 text-match (FR-004)
-  status: ✅ done
-  priority: low
-  source: user
-  area: backend
-  details: CCAP MVP3-candidate — exact-identifier matchre (UserController, REQ-RAG-MGMT-001) BM25 felülmúlja a pure cosine-t. Implementálva: LVS_Search_Mode.hybrid enum value + vectorSearch signature bővítés `hybridWeight?: {vector,text}` (default 0.5/0.5) + `textSearchKey: keyof T` (required hybrid módban). Pure-TS DyNTS_LVS_BM25_Corpus util (canonical k1=1.2 b=0.75, [\w]+ tokenizer case-insens, BM25+ IDF formula). Hybrid pipeline: cosine ALL candidate-re + BM25 ALL candidate-re + min-max [0,1] norm + weighted linear sum + sort desc + top-N slice. Throws: VS4 (missing textSearchKey) + VS5 (invalid hybridWeight). All-zero BM25 → effective cosine-only fallback. Non-breaking: új enum value + új optional params. Specs: 20/20 BM25 util + 8/8 hybrid (1232 spec total / 1 pre-existing FR-002 unrelated failure). Scope-out: MongoDB Atlas $search DB-backed variant (külön ticket), L2+hybrid kombinacio (explicit reject).
-
-- [FEATURE] (BL-20260517-001) Generic compareData() method a DyNTS_DataService base-en (FR-001)
-  status: ✅ done
-  priority: high
-  source: user
-  area: backend
-  details: CCAP RAG-blokkoló (MP3-B, MP5-A/D/E/F) — eddig csak DyNTS_LVS_DocChunkDataService.compareChunk volt DocChunk-T-specifikus delta-detection; non-DocChunk RAG-okra (Rules/Knowledge/CodingPatterns/Memory) anti-DRY. Implementálva: `compareData(newData, oldData, options?): { result: 'equal'|'modified', changedFields? }` a base-en (sync), default deep-equal (primitive/Date/Array/POJO + Object.is fallback), DyFM_Metadata skip (_id, __created, __createdBy, __lastModified, __lastModifiedBy), `options.fields` scope-szűkít (+ skip-list-override), `options.customComparators` per-field, throw 400 (DyNTS-DS0-CD1) null input + 400 (DyNTS-DS0-CD2) üres fields. Non-breaking: új optional method; meglévő compareChunk változatlan. Spec: __agent/feature-requests/FR-001-generic-compareData.md (workspace). 15/15 új jasmine spec + 1195/0 full suite zöld.
-
-- [FEATURE] (BL-20260417-001) Dynamo NTS kliens kiszolgálás képesség
-  status: ✅ done
-  priority: medium
-  source: user
-  area: backend
-  details: Alkalmassá kell tenni a Dynamo NTS-t, hogy kiszolgálja a klienst (frontend/client-side serving). Audit eredménye (2026-05-17): a capability MÁR implementálva van 2026-03-30 óta (commit 8ffe535) — DyNTS_App_Server.mountStaticClient() + getStaticClientSettings?() opt-in override + DyNTS_StaticClient_Settings interface (root, fallbackPath, assetCacheMaxAge, assetCacheImmutable, fallbackCacheMaxAge). Featureset: Express.static mount /-en az API route-ok után, SPA fallback sendFile-lal, asset+fallback Cache-Control (Angular hashed fájl optimalizálásra), default 404 HTML ha nincs fallbackPath, mindkét http/https express-en mount. Integration test megvan (app-extended.integration.spec.ts + app-integration-test.mock.ts). How-to doc: __documentations/2026-05-17-static-client-serving-howto.md (config matrix + example subclass + cache stratégia + edge case-ek).
-
-- [FEATURE] (BL-20260420-001) Átfogó file-based log kezelési rendszer
-  status: ✅ done
-  priority: medium
-  source: user
-  area: backend
-  details: Config-vezérelt log rendszer, ahol egy boolean flag-gel be lehet kapcsolni a file-ba történő log írást, és további config opciókkal lehet állítani a maximum tárolt log mennyiséget (file size / rotation / retention) és egyéb log kezelési paramétereket. Implementálva (2026-05-17): DyNTS_FileLog_Service (singleton) — process.stdout/stderr.write monkey-patch, per-session log fájl, sync appendFileSync (crash-safe), config: file_log.{enabled, logDir, filenamePrefix, maxFileSizeMb, maxFiles, retentionDays, stripAnsi, includeStdout, includeStderr}. Rotation size-limitre, retention by count + age (ami előbb teljesül). Silent failure (file IO sosem blokkolja a szervert). Spec: 11 spec / 0 fail. Full suite: 1133/0. Smoke verified.
-
-- [FEATURE] (BL-20260420-002) Admin API key alapú authentikációs réteg (env-ből)
-  status: ✅ done
-  priority: high
-  source: user
-  area: backend
-  details: Új admin authentikációs mechanizmus bevezetése, ahol az admin API key-t environment variable-ből olvassuk (pl. DYNTS_ADMIN_API_KEY), és ezzel lehet védeni az érzékeny / admin endpointokat. Implementálva (2026-05-17): DyNTS_AdminApiKey_AuthService (singleton) — preProcess `.verify(req, res)` fn opt-in átadható `DyNTS_Endpoint_Params.preProcesses`-be vagy `DyNTS_getLogsRoutingModule({ authPreProcess })`-be. Default env var DYNTS_ADMIN_API_KEY, header x-admin-api-key + Authorization Bearer fallback. Konfig: `configure({ envVarName, headerName, allowAuthorizationBearer })`. Timing-safe compare (crypto.timingSafeEqual + length-mismatch dummy compare). Hibák: 500 ha env nincs, 401 ha header hiányzik/rossz. Env minden híváskor olvasott (nem cache-elt). 15/15 spec + 1148/0 full suite + smoke verifikálva. Foundational a BL-003 (log fetch) és BL-004 (errors retrofit) entry-knek.
-
-- [FEATURE] (BL-20260420-003) Server log file-ok lekérése admin endpointon keresztül
-  status: ✅ done
-  priority: medium
-  source: user
-  area: backend
-  details: A BL-20260420-001 file-based log rendszerre épülve egy új endpoint, amin keresztül le lehet kérni a szerverről a kiírt log file-ok tartalmát (pl. paged / range / tail módon). Implementálva (2026-05-17): DyNTS_FileLogs_Controller (GET /list + GET /file/:filename) + DyNTS_getFileLogsRoutingModule() factory + DyNTS_FileLog_Service bővítés (listLogFiles, readLogFile, getCurrentLogFilename). Olvasási módok: tail / head / range (max 10000 sor). Safeguards: filename whitelist regex prefix-szel, path traversal védelem (resolvedPath + sep ellenőrzés), 503/404/400 helyes status code-ok. Auth opt-in: `DyNTS_getFileLogsRoutingModule({ authPreProcess: adminAuth.verify })`. Default route /file-logs. 38/38 FileLog spec (24 service + 14 controller) + 1175/0 full suite + smoke 4/4 verifikálva.
-
-- [IMPROVEMENT] (BL-20260420-004) Errors endpointok opcionális admin API key védelme
-  status: ✅ done
-  priority: medium
-  source: user
-  area: backend
-  details: A BL-20260420-002 admin API key auth réteget opcionálisan rá kell tudni húzni a meglévő DyNTS_Errors_Controller endpointjaira (errors.controller.ts) is, hogy érzékeny error-listázó / kezelő endpointok admin auth mögé kerülhessenek. Implementálva (2026-05-17): static `DyNTS_Errors_Controller.configure({ authPreProcess?, protectedEndpoints? })` — opt-in, default = no auth (backwards compatible). `authPreProcess` jelenléte esetén default mind a 8 endpoint védett, vagy `protectedEndpoints` subset-tel fine-grained kontroll. `getAuthConfig()` + `_resetAuthConfigForTesting()` segéd-fnk-ek. Per-endpoint `preProcesses: this.getPreProcessesFor(name)` thread-elve mind a 8 endpoint-konstrukcióba. 16/16 errors controller spec (11 régi backwards-compat + 5 új retrofit) + 1180/0 full suite + smoke 3/3 (default, full, subset).
+# BACKLOG.md
+
+> STRICT PATTERN (MUST FOLLOW EXACTLY)
+> - Entry header: "- [TYPE] (BL-YYYYMMDD-###) TITLE"
+> - Immediately followed by (exact keys, colon-space, indented 2 spaces):
+>   - "  status: <EMOJI status-key>"
+>   - "  priority: <low|medium|high|urgent>"
+>   - "  source: <user|system|assistant>"
+>   - "  area: <ui|backend|infra|docs|tests|ux|general>"
+>   - "  details: <single-line summary>"
+> - NO blank line between header and fields (parser reads i+1..i+5).
+> - One blank line between entries (readability).
+> - TYPE values: FEATURE, BUG, IMPROVEMENT, QUESTION, RESEARCH, TASK
+> - STATUS EMOJI: ❌ ⏳ 🔄 ✅ ⚠️ ❓
+> - IDs: unique, format BL-YYYYMMDD-### (increment ###).
+
+- [IMPROVEMENT] (BL-20260518-003) LVS hybrid: BM25 corpus + numberOfCandidates pre-filter cache (FR-004 followup)
+  status: ⏳ pending
+  priority: low
+  source: assistant
+  area: backend
+  details: A jelenlegi hybrid search minden vectorSearch hívásnál ujraepiti a BM25 corpus-t (O(N×|doc|)) és a cosine-t ALL candidate-re futtatja, nem csak top-N-re. ~100..10000 doc LVS-pool-ra elfogadható (ms-mp), de 100K+ doc + ~1KB/doc → ~100MB token allocation per query + GC pressure. Two-pronged optim: (1) opt-in BM25 index cache (corpus-key alapján, invalidate on data write), (2) `numberOfCandidates` érdemi haszna — cosine top-numberOfCandidates pre-filter, BM25 csak ezeken fut. CCAP RAG batch-hívásokra különösen releváns. Spec: lvs-bm25.util.ts (cache layer) + lvs-local-vector-search.data-service.ts (pre-filter branch hybrid módban).
+
+- [IMPROVEMENT] (BL-20260518-002) LVS hybrid: Partial<hybridWeight> API ergonomics (FR-004 followup)
+  status: ⏳ pending
+  priority: low
+  source: assistant
+  area: backend
+  details: A jelenlegi `vectorSearch({ ..., hybridWeight: { vector: number; text: number } })` típus mindkét mezőt kötelezővé teszi ha `hybridWeight` megadva. Ergonomikusabb: `hybridWeight?: Partial<{ vector: number; text: number }>` — `{ vector: 0.8 }` esetén `text` default 0.5 (vagy `1 - vector` semantic-szal). Default-merge a `vectorSearch`-ben: `wVector = hybridWeight?.vector ?? 0.5`, `wText = hybridWeight?.text ?? (1 - wVector)`. Non-breaking (jelenlegi `{vector:0.7,text:0.3}` továbbra is valid). 1-2 új spec a partial override-ra. Fájl: lvs-local-vector-search.data-service.ts:130 + spec extension.
+
+- [BUG] (BL-20260518-001) FR-002 cost-event batch test mock/assertion mismatch
+  status: ✅ done
+  priority: medium
+  source: assistant
+  area: tests
+  details: Fix opció (a) választva — mockEmbeddingsCreate.and.returnValue per-test override 2-elem data + usage:{prompt_tokens:10,total_tokens:10} a batch szándékát tükrözve. A spec-fájl `it('| createEmbeddings emits 1 cost-event with embedding-batch callType')` (oai-embedding.control-service.spec.ts:~275-290) override-ot kapott a top-level beforeEach (line 22-47) `prompt_tokens:5` default fölé. Verifikáció: `npx tsc` (build regenerate) + `npx jasmine` → 1232 specs / 0 failures. **Process root-cause azonosítva**: `spec/support/jasmine.json` `spec_dir: build/` — a jasmine a TS-fordított JS-ből fut, NEM a src-ből. A 08defee pre-commit smoke-ban `npx tsc --noEmit` (csak type-check, NEM emit) + `npx jasmine` futott → a `build/` STALE volt (az új teszt-em NEM volt benne), így a régi 1195 spec-pool fut zöldön, az új 10 spec-em sem assertelt. **5-stage cycle Test stage policy** (saját magamra): pre-commit jasmine futtatás MINDIG `npx tsc` (emit) ELŐZZE MEG, NEM `--noEmit`-tel ellenőrizni. Followup ez a process-tanulság a memóriába (feedback_test_locally_first kiegészítés).
+
+- [FEATURE] (BL-20260517-002) Per-call cost-event hook on AI services (FR-002)
+  status: ✅ done
+  priority: high
+  source: user
+  area: backend
+  details: CCAP CT_ cost-tracking blokkoló (MP3-C, MP4-D, MP6-D). Non-breaking, additív: új `DyNTS_AI_CostEvent` interface (callType, provider, model, tokensUsed: {input, output?, total}, durationMs, issuer, timestamp, estimatedCostUsd?) + `DyNTS_AI_CostEventCallback` type alias. `DyNTS_AI_Provider_ServiceBase`-en protected `onCostEvent?` field + protected `emitCostEvent(event)` safe helper (try/catch, log warn — callback hibája NEM akasztja meg az AI-call-t). OAI emit-helyek: `DyNTS_OAI_Embedding_ControlService.createEmbedding` ('embedding-single'), `.createEmbeddings` ('embedding-batch'), `DyNTS_OAI_LLM_ServiceBase.resolveMessage` ('llm-completion' ha 1 user-msg, 'llm-chat' ha több), `DyNTS_OAI_LLM_Chat_ServiceBase.resolveConversation` ('llm-chat'). Timing: Date.now() start/end, durationMs >= 0. Provider/model auto-resolved (this.aiProvider + String(callInput.model)). Constructor opt: `DyFM_OAI_Settings & { onCostEvent? }` intersection. `estimatedCostUsd` Dynamo-szinten mindig undefined (nincs pricing registry); CCAP CT_ consumer tölti ki. 'llm-tool-use' callType jövőbeli FR. Anthropic / FDP-AI provider-ek csak az alap-osztály onCostEvent mezőt öröklik (még nem implementálnak embedding/LLM-et, saját emit-helyek külön FR). oai-vector-data.service.ts SKIPPED — fájl nem létezik (a spec hivatkozása konceptuális). 10 új jasmine teszt (4 base + 4 embedding + 2 LLM + 1 constructor). Full suite: 1195/0. Spec: __agent/feature-requests/FR-002-per-call-cost-event-hook.md (workspace).
+
+- [FEATURE] (BL-20260517-003) LVS hybrid search: cosine + BM25 text-match (FR-004)
+  status: ✅ done
+  priority: low
+  source: user
+  area: backend
+  details: CCAP MVP3-candidate — exact-identifier matchre (UserController, REQ-RAG-MGMT-001) BM25 felülmúlja a pure cosine-t. Implementálva: LVS_Search_Mode.hybrid enum value + vectorSearch signature bővítés `hybridWeight?: {vector,text}` (default 0.5/0.5) + `textSearchKey: keyof T` (required hybrid módban). Pure-TS DyNTS_LVS_BM25_Corpus util (canonical k1=1.2 b=0.75, [\w]+ tokenizer case-insens, BM25+ IDF formula). Hybrid pipeline: cosine ALL candidate-re + BM25 ALL candidate-re + min-max [0,1] norm + weighted linear sum + sort desc + top-N slice. Throws: VS4 (missing textSearchKey) + VS5 (invalid hybridWeight). All-zero BM25 → effective cosine-only fallback. Non-breaking: új enum value + új optional params. Specs: 20/20 BM25 util + 8/8 hybrid (1232 spec total / 1 pre-existing FR-002 unrelated failure). Scope-out: MongoDB Atlas $search DB-backed variant (külön ticket), L2+hybrid kombinacio (explicit reject).
+
+- [FEATURE] (BL-20260517-001) Generic compareData() method a DyNTS_DataService base-en (FR-001)
+  status: ✅ done
+  priority: high
+  source: user
+  area: backend
+  details: CCAP RAG-blokkoló (MP3-B, MP5-A/D/E/F) — eddig csak DyNTS_LVS_DocChunkDataService.compareChunk volt DocChunk-T-specifikus delta-detection; non-DocChunk RAG-okra (Rules/Knowledge/CodingPatterns/Memory) anti-DRY. Implementálva: `compareData(newData, oldData, options?): { result: 'equal'|'modified', changedFields? }` a base-en (sync), default deep-equal (primitive/Date/Array/POJO + Object.is fallback), DyFM_Metadata skip (_id, __created, __createdBy, __lastModified, __lastModifiedBy), `options.fields` scope-szűkít (+ skip-list-override), `options.customComparators` per-field, throw 400 (DyNTS-DS0-CD1) null input + 400 (DyNTS-DS0-CD2) üres fields. Non-breaking: új optional method; meglévő compareChunk változatlan. Spec: __agent/feature-requests/FR-001-generic-compareData.md (workspace). 15/15 új jasmine spec + 1195/0 full suite zöld.
+
+- [FEATURE] (BL-20260417-001) Dynamo NTS kliens kiszolgálás képesség
+  status: ✅ done
+  priority: medium
+  source: user
+  area: backend
+  details: Alkalmassá kell tenni a Dynamo NTS-t, hogy kiszolgálja a klienst (frontend/client-side serving). Audit eredménye (2026-05-17): a capability MÁR implementálva van 2026-03-30 óta (commit 8ffe535) — DyNTS_App_Server.mountStaticClient() + getStaticClientSettings?() opt-in override + DyNTS_StaticClient_Settings interface (root, fallbackPath, assetCacheMaxAge, assetCacheImmutable, fallbackCacheMaxAge). Featureset: Express.static mount /-en az API route-ok után, SPA fallback sendFile-lal, asset+fallback Cache-Control (Angular hashed fájl optimalizálásra), default 404 HTML ha nincs fallbackPath, mindkét http/https express-en mount. Integration test megvan (app-extended.integration.spec.ts + app-integration-test.mock.ts). How-to doc: __documentations/2026-05-17-static-client-serving-howto.md (config matrix + example subclass + cache stratégia + edge case-ek).
+
+- [FEATURE] (BL-20260420-001) Átfogó file-based log kezelési rendszer
+  status: ✅ done
+  priority: medium
+  source: user
+  area: backend
+  details: Config-vezérelt log rendszer, ahol egy boolean flag-gel be lehet kapcsolni a file-ba történő log írást, és további config opciókkal lehet állítani a maximum tárolt log mennyiséget (file size / rotation / retention) és egyéb log kezelési paramétereket. Implementálva (2026-05-17): DyNTS_FileLog_Service (singleton) — process.stdout/stderr.write monkey-patch, per-session log fájl, sync appendFileSync (crash-safe), config: file_log.{enabled, logDir, filenamePrefix, maxFileSizeMb, maxFiles, retentionDays, stripAnsi, includeStdout, includeStderr}. Rotation size-limitre, retention by count + age (ami előbb teljesül). Silent failure (file IO sosem blokkolja a szervert). Spec: 11 spec / 0 fail. Full suite: 1133/0. Smoke verified.
+
+- [FEATURE] (BL-20260420-002) Admin API key alapú authentikációs réteg (env-ből)
+  status: ✅ done
+  priority: high
+  source: user
+  area: backend
+  details: Új admin authentikációs mechanizmus bevezetése, ahol az admin API key-t environment variable-ből olvassuk (pl. DYNTS_ADMIN_API_KEY), és ezzel lehet védeni az érzékeny / admin endpointokat. Implementálva (2026-05-17): DyNTS_AdminApiKey_AuthService (singleton) — preProcess `.verify(req, res)` fn opt-in átadható `DyNTS_Endpoint_Params.preProcesses`-be vagy `DyNTS_getLogsRoutingModule({ authPreProcess })`-be. Default env var DYNTS_ADMIN_API_KEY, header x-admin-api-key + Authorization Bearer fallback. Konfig: `configure({ envVarName, headerName, allowAuthorizationBearer })`. Timing-safe compare (crypto.timingSafeEqual + length-mismatch dummy compare). Hibák: 500 ha env nincs, 401 ha header hiányzik/rossz. Env minden híváskor olvasott (nem cache-elt). 15/15 spec + 1148/0 full suite + smoke verifikálva. Foundational a BL-003 (log fetch) és BL-004 (errors retrofit) entry-knek.
+
+- [FEATURE] (BL-20260420-003) Server log file-ok lekérése admin endpointon keresztül
+  status: ✅ done
+  priority: medium
+  source: user
+  area: backend
+  details: A BL-20260420-001 file-based log rendszerre épülve egy új endpoint, amin keresztül le lehet kérni a szerverről a kiírt log file-ok tartalmát (pl. paged / range / tail módon). Implementálva (2026-05-17): DyNTS_FileLogs_Controller (GET /list + GET /file/:filename) + DyNTS_getFileLogsRoutingModule() factory + DyNTS_FileLog_Service bővítés (listLogFiles, readLogFile, getCurrentLogFilename). Olvasási módok: tail / head / range (max 10000 sor). Safeguards: filename whitelist regex prefix-szel, path traversal védelem (resolvedPath + sep ellenőrzés), 503/404/400 helyes status code-ok. Auth opt-in: `DyNTS_getFileLogsRoutingModule({ authPreProcess: adminAuth.verify })`. Default route /file-logs. 38/38 FileLog spec (24 service + 14 controller) + 1175/0 full suite + smoke 4/4 verifikálva.
+
+- [IMPROVEMENT] (BL-20260420-004) Errors endpointok opcionális admin API key védelme
+  status: ✅ done
+  priority: medium
+  source: user
+  area: backend
+  details: A BL-20260420-002 admin API key auth réteget opcionálisan rá kell tudni húzni a meglévő DyNTS_Errors_Controller endpointjaira (errors.controller.ts) is, hogy érzékeny error-listázó / kezelő endpointok admin auth mögé kerülhessenek. Implementálva (2026-05-17): static `DyNTS_Errors_Controller.configure({ authPreProcess?, protectedEndpoints? })` — opt-in, default = no auth (backwards compatible). `authPreProcess` jelenléte esetén default mind a 8 endpoint védett, vagy `protectedEndpoints` subset-tel fine-grained kontroll. `getAuthConfig()` + `_resetAuthConfigForTesting()` segéd-fnk-ek. Per-endpoint `preProcesses: this.getPreProcessesFor(name)` thread-elve mind a 8 endpoint-konstrukcióba. 16/16 errors controller spec (11 régi backwards-compat + 5 új retrofit) + 1180/0 full suite + smoke 3/3 (default, full, subset).

package/_specifications/TODO.md

@@ -1,15 +1,15 @@
-# TODO.md
-
-> STRICT PATTERN (MUST FOLLOW EXACTLY)
-> - Entry header: "- EMOJI (TD-YYYYMMDD-###) TITLE"
-> - Immediately followed by (exact keys, colon-space, indented 2 spaces):
->   - "  priority: <low|medium|high|urgent>"
->   - "  area: <ui|backend|infra|docs|tests|ux|general>"
->   - "  source: <user|system|assistant|spec-file.md>"
->   - "  details: <single-line description>"
-> - NO blank line between header and fields (parser reads i+1..i+4).
-> - One blank line between entries (readability).
-> - EMOJI values: ❌ ⏳ 🔄 🔍 ✅ ⚠️ ❓
-> - IDs: unique per project, format TD-YYYYMMDD-### (increment ###).
-> - Group entries by ## headings (spec file / feature area).
-> - One deliverable per entry (atomic tasks).
+# TODO.md
+
+> STRICT PATTERN (MUST FOLLOW EXACTLY)
+> - Entry header: "- EMOJI (TD-YYYYMMDD-###) TITLE"
+> - Immediately followed by (exact keys, colon-space, indented 2 spaces):
+>   - "  priority: <low|medium|high|urgent>"
+>   - "  area: <ui|backend|infra|docs|tests|ux|general>"
+>   - "  source: <user|system|assistant|spec-file.md>"
+>   - "  details: <single-line description>"
+> - NO blank line between header and fields (parser reads i+1..i+4).
+> - One blank line between entries (readability).
+> - EMOJI values: ❌ ⏳ 🔄 🔍 ✅ ⚠️ ❓
+> - IDs: unique per project, format TD-YYYYMMDD-### (increment ###).
+> - Group entries by ## headings (spec file / feature area).
+> - One deliverable per entry (atomic tasks).