@exellix/ai-tasks 9.1.0 → 10.0.0

package/documenations/sessions/2026-06-08-subnets-model-resolution/INVESTIGATION-original-bug.md

@@ -1,182 +0,0 @@
-# Investigation: `Unknown model "google/gemini-2.5-flash-lite"`
-
-Status: **runtime / consumer deps** (skill catalog ruled out)  
-Filed: 2026-06-08  
-Repro: `worox-graphs-playground` — `graph:subnets-analysis`, `node:inbound-reachability`, `skillKey=professional-answer`  
-Verified: 2026-06-08 — **no models in skill/funcx content catalog**; runtime `cheap/default` → wire slug is correct.
-
----
-
-## Executive summary
-
-| Question | Answer |
-|----------|--------|
-| Was `google/gemini-2.5-flash-lite` the authored input? | **No** — runtime uses `cheap/default`. |
-| Is `cheap/default` → `google/gemini-2.5-flash-lite` correct? | **Yes** — expected OpenRouter slug from `@x12i/ai-profiles` at **invoke time**. |
-| Did it come from the skill catalog? | **No** — Catalox skill rows store no model; verified. |
-| Which phase failed? | Almost certainly **PRE or POST xynthesis** (funcx), **not** MAIN `professional-answer` (ai-gateway). |
-| Why does the error say `skillKey=professional-answer`? | Graph-engine wraps the whole `runTask` failure under the node task key. |
-
-**Conclusion:** Runtime profile resolution is fine. If the error persists, it is **funcx + `@x12i/ai-tools` package catalog lookup** at invoke (or misleading phase in the error wrapper) — **not** skill catalog data and not wrong input from `cheap/default`.
-
-### “Catalog” disambiguation
-
-| Name | Role in this incident |
-|------|----------------------|
-| Catalox **ai-skills** catalog | Templates + Optimixer only — **not involved** |
-| **Runtime** `modelConfig` / graph plan | Source of `cheap/default` — **correct** |
-| `@x12i/ai-tools` **models catalog** (inside funcx) | Validates `google/gemini-2.5-flash-lite` at `ask()` — **only place “catalog-resolvable” applies** |
-
----
-
-## Observed error
-
-```json
-{
-  "nodeId": "node:inbound-reachability",
-  "error": {
-    "message": "runTask failed (skillKey=professional-answer nodeId=node:inbound-reachability graphId=graph:subnets-analysis): Unknown model \"google/gemini-2.5-flash-lite\". Pass a catalog-resolvable model slug (e.g. \"deepseek/deepseek-v4-flash\")."
-  }
-}
-```
-
-Stack: `@exellix/graph-engine` → `runTask` → root error from `@x12i/funcx` `requireCatalogModel()`.
-
-Exact funcx message shape (from `@x12i/funcx` ≥4.3):
-
-```text
-Unknown model "<slug>". Pass a catalog-resolvable model slug (e.g. "deepseek/deepseek-v4-flash").
-```
-
-ai-gateway / ai-tools `ModelResolutionError` uses a **different** message (`Could not resolve model "..." (provider: "...")`). So this failure is **not** the MAIN gateway path.
-
----
-
-## Resolution chain (verified in mono-repo)
-
-### 1. ai-profiles: `cheap/default`
-
-Bundled registry (`@x12i/ai-profiles`):
-
-- Profile `cheap`, choice `default`
-- `provider: google`, `modelId: gemini-2.5-flash-lite`
-- OpenRouter invocation: `google/gemini-2.5-flash-lite`
-
-### 2. xynthesis: alias → bare OpenRouter slug
-
-`resolveXynthesisModel("cheap/default")` → `wireModelId: "google/gemini-2.5-flash-lite"`  
-Used by `FuncxInvoker.client.ask({ model: wireModelId })`.
-
-### 3. ai-tasks PRE step
-
-When `taskConfiguration.aiTaskStrategies.pre` includes synthesis, `_runSynthesizedContextPreStep` reads `modelConfig.preActionModel` (typically `cheap/default` for subnet-style graphs).
-
-Fixture pattern in this repo:
-
-```json
-"modelConfig": {
-  "preActionModel": "cheap/default",
-  "postActionModel": "cheap/default",
-  "skillModel": "<MAIN profile or concrete — separate from PRE>"
-}
-```
-
-### 4. MAIN professional-answer (different path)
-
-`skillModel` resolves to **gateway** wire id, e.g. `openrouter/google/gemini-2.5-flash-lite` — not the bare slug in the error.
-
----
-
-## Failure hypothesis (ranked)
-
-### H1 — Stale `@x12i/ai-tools` / `@x12i/funcx` in `worox-graphs-playground` (most likely)
-
-funcx `requireCatalogModel()` calls ai-tools `createModelNameResolver(catalog).resolve()`. If the installed catalog predates `gemini-2.5-flash-lite`, or OpenRouter lane index is incomplete, `found: false` → this exact error.
-
-**Mono-repo reference versions (known good):**
-
-| Package | Version in `ai-tasks` |
-|---------|----------------------|
-| `@exellix/ai-tasks` | 9.0.3 |
-| `@exellix/xynthesis` | 4.6.4 |
-| `@x12i/funcx` (override) | ^4.6.0 |
-| `@x12i/ai-profiles` | ^3.3.0 |
-
-### H2 — funcx ai-tools client disabled or throws in catalog load
-
-funcx `resolveModelSlug` catches all errors and returns `found: false` — surfaces as "Unknown model" even when the slug is valid.
-
-Check env / config: FuncX ai-tools `enabled=false`, broken catalog fetch, etc.
-
-### H3 — Compile-time concrete slug injection (secondary)
-
-If graph plan has `resolvedInvocationSnapshot.modelId: "google/gemini-2.5-flash-lite"` on a PRE unit, behavior is the same at funcx — but **authoring** was still `cheap/default`. See CR-1.
-
-### H4 — Wrong phase in operator UI (symptom only)
-
-Operator reads "professional-answer failed" and assumes MAIN model is wrong. PRE failed before MAIN ran. See FR-2.
-
----
-
-## Investigation checklist (`worox-graphs-playground`)
-
-Run in the consumer repo (not this mono-repo unless linked):
-
-- [ ] **1. Confirm runtime `modelConfig` on the failing node**
-  - `preActionModel`, `skillModel`, `postActionModel` on the `RunTaskRequest` graph-engine builds.
-  - Expect: `cheap/default` on PRE; MAIN slot may differ.
-
-- [ ] **2. Confirm compiled `nodePlan` for `node:inbound-reachability`**
-  - Per-unit `modelSelection` (`profileChoice` vs concrete).
-  - `resolvedInvocationSnapshot.modelId` — if present with concrete slug, note for CR-1.
-
-- [ ] **3. Lockfile versions**
-  ```bash
-  npm ls @x12i/funcx @x12i/ai-tools @exellix/xynthesis @exellix/ai-tasks @x12i/ai-profiles
-  ```
-  Compare to mono-repo `ai-tasks/package.json`.
-
-- [ ] **4. Catalog spot-check in installed tree**
-  ```bash
-  rg "gemini-2.5-flash-lite" node_modules/@x12i/ai-tools/dist/
-  ```
-  Should appear in bundled `models-catalog.json` chunk.
-
-- [ ] **5. Reproduce with trace**
-  - `executionMode: "trace"` on `runTask`.
-  - Look for xynthesis `wireModelId` / `profileChoiceKey: "cheap/default"` **before** any MAIN skill observation.
-  - If only PRE logs appear → confirms H4.
-
-- [ ] **6. Minimal isolated PRE call**
-  - xynthesis `executeXynthesisAction` with `model: "cheap/default"` and same env as playground.
-  - If fails → H1/H2. If passes → graph wiring or plan overlay issue.
-
-- [ ] **7. OPENROUTER_API_KEY**
-  - Required for OpenRouter backend; missing key causes other errors, but verify present.
-
----
-
-## What is **not** the bug
-
-- Choosing `cheap/default` for PRE/POST.
-- ai-profiles mapping `cheap/default` → `google/gemini-2.5-flash-lite`.
-- xynthesis emitting bare OpenRouter slug to funcx (by design since xynthesis ≥4.1.8).
-- Skill / funcx **content** catalog storing or supplying a model (verified: **no models there**).
-- Confusing runtime resolved wire with catalog-authored model (there is no latter).
-
----
-
-## Recommended fix path (consumer)
-
-1. Align `@x12i/funcx`, `@x12i/ai-tools`, `@exellix/xynthesis`, `@exellix/ai-tasks` to mono-repo versions (or newer).
-2. Re-run graph; if still failing, run checklist §5–6.
-3. If catalog is current and resolve still fails, file upstream bug on funcx/ai-tools with `resolveModelSlug` diagnostics (`attemptedStrategies`, `bestRejectedCandidate`).
-
----
-
-## Related mono-repo references
-
-- PRE fixture: `ai-tasks/test/fixtures/run-task/graph-engine-pre-synthesis.json`
-- xynthesis wire: `xynthesis/src/funcxInvoker.ts`, `xynthesis/src/resolveAiProfileModel.ts`
-- funcx validation: `node_modules/@x12i/funcx/dist/src/serve.js` → `requireCatalogModel`
-- Error wrap: `ai-tasks/src/errors/runTaskExecutionError.ts`

package/documenations/sessions/2026-06-08-subnets-model-resolution/PROBLEM.md

@@ -1,236 +0,0 @@
-# Problem knowledge: `Unknown model "google/gemini-2.5-flash-lite"`
-
-**Session:** 2026-06-08  
-**Repro:** `worox-graphs-playground` — `graph:subnets-analysis`, `node:inbound-reachability`, `skillKey=professional-answer`  
-**Status:** **Resolved** in **`@x12i/funcx@4.9.5+`** (catalog gate via **`@x12i/ai-profiles`**). Ship **`@exellix/xynthesis@4.6.6+`** and **`@exellix/ai-tasks@9.0.5+`** with **`@x12i/funcx@^4.9.5`** declared in dependencies.
-
----
-
-## 1. What we observed
-
-```json
-{
-  "nodeId": "node:inbound-reachability",
-  "error": {
-    "message": "runTask failed (skillKey=professional-answer nodeId=node:inbound-reachability graphId=graph:subnets-analysis): Unknown model \"google/gemini-2.5-flash-lite\". Pass a catalog-resolvable model slug (e.g. \"deepseek/deepseek-v4-flash\")."
-  }
-}
-```
-
-Stack (from consumer): `@exellix/graph-engine` → `@exellix/ai-tasks` `runTask` → … → `@x12i/funcx` `Client.ask()`.
-
-The exact substring `Pass a catalog-resolvable model slug` is thrown only by **`@x12i/funcx`** `requireCatalogModel()` (funcx ≥4.3). It is **not** from ai-gateway, ai-skills, or Catalox.
-
----
-
-## 2. What we verified about input (runtime data)
-
-| Fact | Source |
-|------|--------|
-| Runtime model input was **`cheap/default`**, not a concrete slug in authoring | Operator check |
-| **Skill catalog (Catalox `ai-skills`) stores no model** for `professional-answer` or other skills — templates + Optimixer only | Operator check + mono-repo `AI_SKILLS_CATALOG_ITEMS` |
-| `cheap/default` → `google/gemini-2.5-flash-lite` is **correct** OpenRouter wire output for PRE/POST xynthesis | `@x12i/ai-profiles` bundled registry |
-
-**Important:** The error string shows a **resolved wire slug**, not what was typed in the graph. That slug is **invoke-time output**, not catalog-authored data.
-
----
-
-## 3. Three different “catalog” meanings (do not mix)
-
-| Name | What it is | Role in this incident |
-|------|------------|------------------------|
-| **Catalox skill catalog** | Skill templates, Optimixer fields | **Not involved** — no models |
-| **Runtime orchestration data** | `RunTaskRequest.modelConfig`, graph `modelSelection`, per-unit slots | **Source of `cheap/default`** — correct |
-| **`@x12i/ai-tools` models catalog** | JSON bundled inside the **npm package**; used by funcx/gateway for slug lookup + pricing | **Where the error is thrown** — funcx asks “is this slug known?” |
-| **`@x12i/ai-profiles` registry** | Profile/choice → vendor SKU + OpenRouter invocation mapping | **Already used successfully** before funcx (see §4) |
-
----
-
-## 4. End-to-end path today (what actually runs)
-
-### 4.1 Graph → ai-tasks
-
-For subnet-style nodes running `professional-answer` with PRE synthesis, a typical `RunTaskRequest` looks like:
-
-```json
-{
-  "skillKey": "professional-answer",
-  "modelConfig": {
-    "preActionModel": "cheap/default",
-    "skillModel": "<profile or concrete — MAIN, separate hop>",
-    "postActionModel": "cheap/default"
-  },
-  "taskConfiguration": {
-    "aiTaskStrategies": { "pre": "synthesis", ... }
-  }
-}
-```
-
-Graph-engine wraps any `runTask` failure with `skillKey=professional-answer` even when **PRE dies before MAIN** — misleading for operators.
-
-### 4.2 PRE synthesis — **first hop: ai-profiles** (already in use)
-
-1. ai-tasks `_runSynthesizedContextPreStep` reads `preActionModel` → `"cheap/default"`.
-2. Calls `@exellix/xynthesis` `executeXynthesisAction({ model: "cheap/default", ... })`.
-3. xynthesis `resolveXynthesisModel("cheap/default")` calls **`@x12i/ai-profiles`**:
-   - Profile `cheap`, choice `default`
-   - Vendor: `google` / `gemini-2.5-flash-lite`
-   - OpenRouter invocation: `google/gemini-2.5-flash-lite`
-4. xynthesis builds `modelResolution` snapshot and calls `FuncxInvoker` with:
-   - `model: "cheap/default"` (alias)
-   - `modelResolution.wireModelId: "google/gemini-2.5-flash-lite"`
-
-If step 3 fails, xynthesis throws **before** funcx — e.g. unknown profile/choice.  
-**We got past step 3** (otherwise the error message would be about alias resolution, not “Unknown model … slug”).
-
-### 4.3 PRE synthesis — **second hop: funcx + ai-tools** (where it breaks)
-
-5. `FuncxInvoker` → `client.ask(userPrompt, { model: "google/gemini-2.5-flash-lite", ... })`.
-6. funcx `resolveAskRuntime()` → `requireCatalogModel({ model, provider: "openrouter" })`.
-7. funcx `resolveModelSlug()` loads **`@x12i/ai-tools`** catalog via `getAiModelsCatalogClient().getAllModels()` and runs `createModelNameResolver(catalog).resolve(...)`.
-8. If `found: false` → throws:
-
-   ```text
-   Unknown model "google/gemini-2.5-flash-lite". Pass a catalog-resolvable model slug ...
-   ```
-
-funcx README (≥4.3) documents this split explicitly:
-
-- **Host (xynthesis):** resolve `profile/choice` with **ai-profiles**, pass **concrete slug** to funcx.
-- **funcx:** slug must be **catalog-resolvable** in **ai-tools** — funcx does **not** resolve `cheap/default` itself.
-
-### 4.4 MAIN `professional-answer` — different path (not this error)
-
-MAIN uses `skillModel` → `@exellix/ai-skills` → `@x12i/ai-gateway`:
-
-- Wire shape: `openrouter/google/gemini-2.5-flash-lite` (gateway prefix), not bare slug.
-- Validation: ai-tools `resolveInvokeModel` via gateway, **different error message** (`Could not resolve model "..." (provider: "...")`).
-
-So this incident is **PRE (or POST) xynthesis/funcx**, not MAIN professional-answer gateway invoke.
-
----
-
-## 5. ai-profiles vs ai-tools — what is done today, could ai-profiles replace ai-tools here?
-
-### What is done today
-
-| Step | Package | Job |
-|------|---------|-----|
-| Profile → wire slug | **ai-profiles** (in xynthesis) | `cheap/default` → `google/gemini-2.5-flash-lite` |
-| Slug accepted for HTTP call? | **ai-tools** (inside funcx) | Re-lookup slug in bundled models JSON |
-| LLM HTTP call | **funcx** → OpenRouter | Uses validated `modelId` |
-
-**ai-profiles is already in the pipeline.** The failure is not “we should use ai-profiles instead of nothing” — it is **ai-tools rejecting a slug that ai-profiles just produced**.
-
-### Could funcx validate with ai-profiles instead of ai-tools?
-
-**Theoretically yes, but not how funcx is built today.**
-
-- ai-profiles exposes `ensureConcreteModel("google/gemini-2.5-flash-lite")` and `resolveCatalogModel(...)` against its own `models-catalog.json` / registry linkage — similar knowledge, **different package and index**.
-- funcx 4.x is wired to **ai-tools** for `requireCatalogModel`, pricing (`readCallCostFromAskResult`), and Activix `autoCost`.
-- funcx README requires both packages aligned (e.g. **^3.3.0**).
-
-**Practical options (no code changed in this session):**
-
-| Approach | Effect |
-|----------|--------|
-| **Align versions** in `worox-graphs-playground`: `@x12i/funcx`, `@x12i/ai-tools`, `@x12i/ai-profiles`, `@exellix/xynthesis` to mono-repo pins | Most likely fix — both catalogs should agree on `gemini-2.5-flash-lite` |
-| **Pass `cheap/default` into funcx `ask()`** | **Fails by design** — funcx requires concrete catalog slug, not profile alias |
-| **funcx skip ai-tools check when `modelResolution` from ai-profiles present** | Would need **funcx product change** — trust host resolution, use ai-tools only for cost |
-| **funcx call ai-profiles `resolveCatalogModel` instead of ai-tools resolver** | Would need **funcx product change** — duplicate validation path, still two catalogs unless unified |
-
-**“Work better with ai-profiles”** in the current architecture means: **keep resolving in xynthesis via ai-profiles (already done), then ensure the ai-tools copy in the same install recognizes the emitted slug** — not replace step 1, fix step 2 alignment.
-
----
-
-## 6. Why ai-profiles can succeed and ai-tools can still fail
-
-Both use bundled JSON that can **drift across package versions** or **resolve differently**:
-
-- ai-profiles: registry + `normalizeInvokeModel` → OpenRouter slug from profile choice.
-- ai-tools: `ModelNameResolver` over merged direct + OpenRouter catalog maps; lookup strategies (exact, slug, provider-prefix, fallback catalog).
-
-Known mono-repo state (reference install):
-
-- **ai-tools** direct catalog includes `provider: google`, `modelId: gemini-2.5-flash-lite`.
-- **ai-tools** OpenRouter catalog may list newer ids (e.g. `google/gemini-3.1-flash-lite`) without listing `2.5-flash-lite` under OpenRouter lane — resolver should fall back to direct catalog, **if consumer ai-tools version is current**.
-
-Failure modes for step 7:
-
-1. **Older `@x12i/ai-tools`** in playground — slug not in bundled JSON.
-2. **Older `@x12i/funcx`** — different resolver behavior.
-3. **ai-tools client disabled / throws** — funcx catches and returns `found: false` → same “Unknown model” message.
-4. **Version skew** — ai-profiles 3.3 emits slug ai-tools 2.x does not index.
-
----
-
-## 7. What we know vs what we do not know
-
-### Known
-
-- Input at runtime: `cheap/default` (verified).
-- Skill Catalox catalog: no models (verified).
-- Wire slug in error is expected output of ai-profiles for `cheap/default`.
-- Error originates in funcx `requireCatalogModel` (ai-tools gate), not skill catalog.
-- Error message shape implies PRE/POST funcx path, not MAIN gateway.
-- graph-engine attributes failure to `skillKey=professional-answer` regardless of phase.
-
-### Not confirmed in consumer repo (if debugging continues)
-
-- Installed versions: `npm ls @x12i/funcx @x12i/ai-tools @x12i/ai-profiles @exellix/xynthesis`
-- Whether error still reproduces after version align.
-- Whether `executionMode: "trace"` shows `profileChoiceKey: "cheap/default"` and `wireModelId` on PRE hop only.
-- Whether `OPENROUTER_API_KEY` is set (missing key usually causes different errors, but worth ruling out).
-
-### Ruled out
-
-- Wrong profile choice in authoring (`cheap/default` is valid).
-- Skill catalog supplying `google/gemini-2.5-flash-lite`.
-- xynthesis double-prefix bug (`openrouter/google/...` to OpenRouter API) — fixed in xynthesis ≥4.1.8; error shows **bare** slug, which is correct for funcx.
-
----
-
-## 8. Minimal reproduction mental model
-
-```text
-cheap/default
-  → [ai-profiles] OK → google/gemini-2.5-flash-lite
-  → [funcx ask + ai-tools requireCatalogModel] FAIL? → Unknown model
-  → [OpenRouter HTTP] never reached
-```
-
-If ai-profiles step failed, you would never see this exact funcx message with that slug.
-
----
-
-## 9. Recommended next step (consumer only)
-
-In `worox-graphs-playground`:
-
-```bash
-npm ls @x12i/funcx @x12i/ai-tools @x12i/ai-profiles @exellix/xynthesis @exellix/ai-tasks
-rg "gemini-2.5-flash-lite" node_modules/@x12i/ai-tools/dist/
-```
-
-Align to mono-repo reference (`ai-tasks/package.json`: funcx ^4.6.0 override, ai-profiles ^3.3.0, xynthesis ^4.6.4). Re-run graph.
-
-If still failing with current catalogs, file upstream on **funcx/ai-tools** with: input alias `cheap/default`, ai-profiles wire `google/gemini-2.5-flash-lite`, and ai-tools resolver `found: false` diagnostics.
-
----
-
-## 10. Mono-repo code references
-
-| Piece | Location |
-|-------|----------|
-| PRE reads `preActionModel` | `ai-tasks/src/core/task-sdk.ts` `_runSynthesizedContextPreStep` |
-| ai-profiles resolve | `xynthesis/src/resolveAiProfileModel.ts` `resolveXynthesisModel` |
-| Pass wire to funcx | `xynthesis/src/executeXynthesisAction.ts`, `xynthesis/src/funcxInvoker.ts` |
-| funcx ai-tools gate | `node_modules/@x12i/funcx/dist/src/serve.js` `requireCatalogModel` |
-| cheap/default → gemini | `node_modules/@x12i/ai-profiles/dist/data/models-profiles.json` |
-| Subnet-style fixture | `ai-tasks/test/fixtures/run-task/graph-engine-pre-synthesis.json` |
-
----
-
-## 11. Session CR/FR docs
-
-Deferred — not part of this problem analysis. See sibling files only if needed for future product work.

package/documenations/sessions/2026-06-08-subnets-model-resolution/README.md

@@ -1,11 +0,0 @@
-# Session: subnets-analysis model resolution (2026-06-08)
-
-**Canonical doc:** [PROBLEM.md](./PROBLEM.md) — everything we know about the `Unknown model "google/gemini-2.5-flash-lite"` failure.
-
-## One-line summary
-
-Runtime input `cheap/default` is correct; ai-profiles resolves it to `google/gemini-2.5-flash-lite`; **funcx then re-validates that slug against `@x12i/ai-tools`** (not the skill catalog) and throws. Skill Catalox catalog has no models.
-
-## Other files in this folder
-
-CR/FR drafts from earlier in the session — **out of scope** for the original bug; kept for reference only.

package/documenations/sessions/2026-06-08-subnets-model-resolution/funcx-test-resolveModel.cheapDefaultWireSlug.test.ts

@@ -1,117 +0,0 @@
-/**
- * Copy to funcx repo: `test/resolveModel.cheapDefaultWireSlug.test.ts`
- *
- * Reproduces PRE xynthesis catalog gate (graphs-studio / ai-tasks):
- *   cheap/default → ai-profiles wire slug → funcx resolveModelSlug / requireCatalogModel
- *
- * Run (from funcx root):
- *   mode=debug FUNCX_ACTIVIX_ENABLED=0 node --import ./test/helpers/testGlobalSetup.ts --import tsx --test test/resolveModel.cheapDefaultWireSlug.test.ts
- *
- * BEFORE fix (funcx <4.9, resolveModelSlug uses getAllModels + narrow resolver):
- *   - ai-tools catalog.resolveModel → found: true
- *   - funcx resolveModelSlug         → found: false  → THIS TEST FAILS
- *
- * AFTER fix (funcx >=4.9, resolveModelSlug delegates to @x12i/ai-profiles requireCatalogModel):
- *   - both → found: true → THIS TEST PASSES
- */
-import assert from "node:assert/strict";
-import { describe, it, beforeEach } from "node:test";
-import { ensureConcreteModel } from "@x12i/ai-profiles";
-import {
-  createAiToolsInvokeClient,
-  createModelNameResolver,
-} from "@x12i/ai-tools";
-import {
-  configureAiToolsEngine,
-  requireCatalogModel,
-  resetAiToolsClients,
-  resolveModelSlug,
-} from "../src/index.js";
-
-const PRE_ALIAS = "cheap/default";
-/** Wire slug xynthesis passes to funcx ask() after ai-profiles resolution. */
-const EXPECTED_WIRE_SLUG = "google/gemini-2.5-flash-lite";
-
-function useDefaultDevCatalogEngine(): void {
-  resetAiToolsClients();
-  configureAiToolsEngine({
-    bundledOnly: false,
-    catalogLane: "text",
-  });
-}
-
-describe("resolveModel — cheap/default PRE wire slug (subnets-analysis repro)", () => {
-  beforeEach(() => {
-    useDefaultDevCatalogEngine();
-  });
-
-  it("step A: ai-profiles resolves cheap/default to the expected OpenRouter wire slug", () => {
-    const resolved = ensureConcreteModel(PRE_ALIAS);
-    assert.equal(
-      resolved.modelId,
-      EXPECTED_WIRE_SLUG,
-      "host/xynthesis must resolve profile alias before funcx"
-    );
-  });
-
-  it("step B–E: funcx must accept the wire slug when ai-tools full resolveModel does (default dev catalog)", async () => {
-    // --- step A (again at runtime boundary) ---
-    const resolved = ensureConcreteModel(PRE_ALIAS);
-    const wireSlug = resolved.modelId;
-    assert.equal(wireSlug, EXPECTED_WIRE_SLUG);
-
-    // --- step C: narrow path input (what funcx uses today) ---
-    const client = createAiToolsInvokeClient({ bundledOnly: false });
-    const mergedMap = await client.catalog.getAllModels();
-    const narrow = createModelNameResolver(mergedMap).resolve({
-      model: wireSlug,
-      provider: "openrouter",
-    });
-
-    // --- step D: ai-tools full path (what funcx SHOULD use) ---
-    const full = await client.catalog.resolveModel({
-      model: wireSlug,
-      provider: "openrouter",
-    });
-
-    // --- step E: funcx public API used by Client.ask() → requireCatalogModel ---
-    const funcx = await resolveModelSlug({
-      model: wireSlug,
-      provider: "openrouter",
-    });
-
-    // Diagnostic log when run with `node --test` (visible in CI output)
-    console.info("[resolveModel.cheapDefaultWireSlug]", {
-      alias: PRE_ALIAS,
-      wireSlug,
-      mergedCatalogSize: mergedMap.size,
-      mergedHasSlug: mergedMap.has(wireSlug),
-      narrowFound: narrow.found,
-      fullResolveFound: full.found,
-      fullResolvedReason: full.found ? full.resolvedReason : full.reason,
-      funcxResolveFound: funcx.found,
-    });
-
-    assert.equal(
-      full.found,
-      true,
-      `ai-tools catalog.resolveModel must accept wire slug "${wireSlug}" (ai-profiles fallback / direct lane)`
-    );
-
-    // Core regression: funcx must not reject a slug that ai-tools full resolver accepts.
-    assert.equal(
-      funcx.found,
-      full.found,
-      "resolveModelSlug must delegate to catalog.resolveModel (not getAllModels narrow map only)"
-    );
-    assert.equal(funcx.modelId, wireSlug);
-
-    // Same gate Client.ask() hits before OpenRouter HTTP:
-    const required = await requireCatalogModel({
-      model: wireSlug,
-      provider: "openrouter",
-    });
-    assert.equal(required.found, true);
-    assert.equal(required.modelId, wireSlug);
-  });
-});

package/documenations/upstream-feature-requests/ai-tasks-wrap-up-after-upstream.md

@@ -1,129 +0,0 @@
-# `@exellix/ai-tasks` — wrap-up after upstream invoke-contract fixes
-
-Status: **unblocked** — **@exellix/xynthesis ≥ 4.3.1** shipped ([xynthesis-ai-profiles-2.1-import-break.md](./xynthesis-ai-profiles-2.1-import-break.md)); invoke-contract items below are shipped in **ai-skills 6.0** / **xynthesis 4.3**:
-
-- [`ai-skills-orchestrator-invoke-contract-5.9.md`](./ai-skills-orchestrator-invoke-contract-5.9.md)
-- [`xynthesis-orchestrator-invoke-contract-4.2.md`](./xynthesis-orchestrator-invoke-contract-4.2.md)
-
-Owner: `@exellix/ai-tasks`  
-Target release: **8.4.x** (or **8.5.0** if breaking `llmCall` surface)
-
----
-
-## Contract (locked with product)
-
-**ai-tasks sends to lower layers:**
-
-- Model ids via `RunTaskModelConfig` triplet (`preActionModel`, `skillModel`, `postActionModel`) + optional `llmCall.model` overrides.
-- `temperature`, `topP`, **`reasoningEffort`** (once upstream exposes it).
-- **`outputExpectation`** on xynthesis-backed hops only (sizing intent — **not** a token count).
-
-**ai-tasks does not send:**
-
-- `maxTokens` / `maxTokensCap` on `modelConfig`, `runSkill`, `executeXynthesisAction`, or direct gateway repair (unless product later revives an explicit ceiling knob with a new name).
-
-**Lower layers own:**
-
-- MAIN caps → `@exellix/ai-skills` Optimixer + Catalox skill catalog.
-- PRE/POST/scoping/utility caps → `@exellix/xynthesis` `resolveEffectiveMaxTokens` inside `executeXynthesisAction`.
-
----
-
-## Already done in ai-tasks (this branch)
-
-- [x] `RunTaskRequest.modelConfig` typed as `RunTaskModelConfig` (not ai-skills `ModelConfig` with `maxTokens`).
-- [x] Removed `maxTokens` from `RunTaskModelConfig`; strip legacy key at ai-skills boundary.
-- [x] Stopped mapping `llmCall.maxTokensCap` → `modelConfig.maxTokens` in `buildAiSkillsModelConfigForMain`.
-- [x] Trace repair: `mapGatewayInvokeToTrace` uses `maxTokensRequested` only (not `modelConfig.maxTokens`).
-- [x] `@exellix/xynthesis` ≥ 4.3.1 in `package.json`; bare profile keys normalized via `toStrictAiProfileResolveInput` before xynthesis invoke.
-- [x] Tests/fixtures use `cheap/default` + `pro/default` (no legacy `balanced` default).
-- [x] Public exports: `listAiTasksProfileChoiceKeys`, `toStrictAiProfileResolveInput`.
-- [x] `npm run build` + `npm run test` green (390 pass / 0 fail).
-
----
-
-## TODO after upstream ships
-
-### 1. Remove broken / duplicate token budgeting
-
-| File / area | Change |
-|-------------|--------|
-| `src/internal/resolveLlmCallForXynthesis.ts` | Remove `resolveMaxTokens` import; either delete token resolution or reduce to **outputExpectation-only** helper (rename e.g. `resolveLlmOutputExpectationForXynthesis`). |
-| `src/internal/runPostStepLlmCall.ts` | Drop `maxTokens` / `resolveLlmCallForXynthesis` pre-resolution; pass **`outputExpectation`** (required). |
-| `src/synthesis/runStructuredSynthesisRobust.ts` | Repair path: stop `resolveMaxTokens`; do not push `maxTokens` into raw AIGateway `config` unless gateway documents a non-Optimixer escape hatch. |
-| `src/index.ts` | Stop re-exporting `resolveMaxTokens`; export `resolveEffectiveMaxTokens` only if graph-engine needs it (prefer not). |
-| `src/utilities/runUtility.ts` | Remove `maxTokens` forward to xynthesis finalize. |
-| `src/post-steps/resolvePostStepConfig.ts` | Remove env `*_MAX_TOKENS_CAP` wiring if contract is zero host caps (or keep env as deprecated no-op one release). |
-
-### 2. Add `reasoningEffort` to ai-tasks surface
-
-| File | Change |
-|------|--------|
-| `src/types/llmCall.ts` | `reasoningEffort?: OptimixerReasoningEffort` (re-export type from `@x12i/optimixer` or xynthesis). |
-| `src/types/model-config.ts` | Optional tuning field; passthrough in `modelConfigTuningPassthrough` (not stripped). |
-| `src/types/task-types.ts` | Document on `llmCall` and post-step configs. |
-| `buildAiSkillsModelConfigForMain` | Forward to ai-skills `modelConfig` when upstream §1 ships. |
-| `runPostStepLlmCall` / `task-sdk` synthesis | Forward to `executeXynthesisAction` / structured params. |
-| `src/validation/helpers.ts` | Optional enum validation if finite set. |
-| `src/builders/task-request-builder.ts` | Document in `withLlmCall` JSDoc. |
-
-### 3. Fix xynthesis call sites (no new upstream needed except §1–2)
-
-| Call site | Pass |
-|-----------|------|
-| `runPostStepLlmCall` → `executeXynthesisAction` | `outputExpectation`, `temperature`, `topP`, `sidekickAction` / `activixActionType`, `reasoningEffort` (when exists) |
-| PRE structured / markdown synthesis (`task-sdk`) | Same via gateway / `runStructuredSynthesisGatewayCall` params |
-| AI scoping | `outputExpectation` (already defaulted in `runScopingCall`) |
-
-### 4. Remove / deprecate `maxTokensCap` on `LlmCallConfig`
-
-- **Breaking:** remove `maxTokensCap` from `LlmCallConfig`, `RunUtilityRequest.exec`, post-step env fallbacks, validation, README.
-- **Or soft:** mark deprecated in 8.4.1, ignore at runtime, remove in 8.5.0.
-
-Confirm with graph-engine before hard removal.
-
-### 5. Tests & docs
-
-- [ ] `npm run build` + `tsc -p tsconfig.test.json` green (no `resolveMaxTokens` import).
-- [ ] `npm run test` green (mocks via `setSynthesisInvoker` unchanged).
-- [ ] Update [`README.md`](../../README.md) token-budget section (remove `llmCall.maxTokensCap` → `modelConfig.maxTokens` overlay story).
-- [ ] Update [`ai-skills-llm-observability.md`](./ai-skills-llm-observability.md) §2 (stale maxTokens wording).
-- [ ] Update [`xynthesis-llm-observability.md`](./xynthesis-llm-observability.md) superseded sections.
-- [ ] Link all three contract docs from [`BREAKING-CHANGES.md`](../../BREAKING-CHANGES.md) 8.4 section.
-
----
-
-## Dependency order
-
-```mermaid
-flowchart LR
-  A[ai-skills: reasoningEffort on RunSkillRequest]
-  B[xynthesis: reasoningEffort on ExecuteXynthesisActionRequest]
-  C[xynthesis: verify temperature topP all paths]
-  D[ai-tasks: remove maxTokens forwarding]
-  E[ai-tasks: wire reasoningEffort + outputExpectation]
-  A --> E
-  B --> E
-  C --> E
-  D --> E
-```
-
-Recommended sequence:
-
-1. Merge **xynthesis** ai-profiles 2.1 import fix (**≥ 4.3.1**) + **ai-skills** contract PRs (can be parallel).
-2. Bump `package.json`: `"@exellix/xynthesis": "^4.3.1"` (minimum — **do not** ship ai-tasks 8.4.x consumer release on **4.3.0**).
-3. Execute **this wrap-up** checklist in one ai-tasks PR.
-4. Release note in **8.4.x** or **8.5.0** with graph-engine migration (drop `maxTokensCap` on task payloads if removed).
-
----
-
-## Quick smoke test after wrap-up
-
-```bash
-npm run build
-npm run test
-# Optional live:
-# npm run test:e2e:synthesis
-```
-
-Expect: no import of `resolveMaxTokens` from `@exellix/xynthesis/ai-actions`; live POST/PRE calls succeed with `outputExpectation` only (no `maxTokens` on request).