pi-freeflow 1.34.0 → 1.36.0

package/CHANGELOG.md

@@ -1,5 +1,64 @@
 # Changelog
 
+## 1.36.0
+
+### Minor Changes
+
+- bf358da: A new free model on the OpenCode Zen lane: Cat Jam
+
+  `cat-jam-free` joins the picker, bringing the catalog to 37 models (11 OpenCode, 15 Kilo, 2 Cline, 9 OVHcloud). No account, no key, no setup — it is keyless like every other Zen free model.
+
+  It thinks, and it has a 262,144-token window. That number was measured the hard way: the gateway happily accepts prompts four times that size and answers normally, but the model can no longer read them. A marker planted at the end of a 248,232-token prompt comes back; at 250,272 tokens it does not. So the advertised window is what the model can actually use, not merely what the service tolerates.
+
+  Two honest limits worth knowing before you pick it:
+
+  - **It is not available on the Responses API.** Requests sent that way are refused with a protocol error. It works on chat completions, which is what hosts use by default, but a host pinned specifically to Responses will not see it.
+  - **Thinking depth starts at "low".** The lowest setting is refused, so "minimal" is not offered; low, medium, high and extra-high all work.
+
+  The output budget is 131,072, the published figure for this model.
+
+### Patch Changes
+
+- bf358da: Long sessions no longer stall on a tool call whose arguments arrived spliced
+
+  Some models open their next tool call by echoing the tail of the one they had just seen, and occasionally stop before the closing brace. The result is a single argument string carrying both faults at once: the previous call's leftover text in front, and the last brace missing. The whole call was then rejected, the turn failed, and the host retried — which resent the same broken payload, so a long session could get stuck repeating it.
+
+  Both faults are now repaired together, so the caller's own arguments arrive intact.
+
+  The repair stays deliberately narrow. It only fires when a leftover fragment is actually present in front of the object, which is the evidence that the model was mid-call rather than that the request itself was malformed, and only when what follows is a complete object merely missing its closing brace. Arguments cut off part-way through a value are still refused untouched, because completing those would silently shorten them — and these arguments carry shell commands.
+
+  Tool calls that were already well-formed are unaffected and pass through byte-identically.
+
+## 1.35.0
+
+### Minor Changes
+
+- 9085408: Nine more free models, from OVHcloud's AI Endpoints — and no account needed
+
+  The picker now offers nine open-weight models served by OVHcloud AI Endpoints, bringing the catalog to 36: GPT-OSS 120B and 20B, four Qwen models (3.5 397B, 3.5 9B, 3.6 27B, 3.8 27B), Mistral Small 3.2 and Mistral 7B, and Meta Llama 3.3 70B. The four Qwens and Mistral Small 3.2 also read images.
+
+  Access is anonymous and keyless. There is nothing to sign up for and no key to paste — no login step, no saved credential, no setup of any kind. Pick a model and it works.
+
+  The honest caveat: OVH rate-limits anonymous callers per IP address. A burst of parallel subagents can run into `API rate limit exceeded`, the same way a shared OpenCode or Kilo quota does. Wait it out or spread the work across models. Requests to these models always go straight to OVH from your machine, never through your relay pool — there is no per-IP gate for a relay to route around.
+
+  Thinking depth is limited to the two levels OVH confirms, `low` and `high`, on the six models that reason. The output budgets are deliberately conservative: they are the smallest cap any provider offers for the same underlying model, so a turn never asks for more than OVH will give.
+
+### Patch Changes
+
+- 3408dad: Stop conversations from dying on request shapes the provider rejects
+
+  Three shapes that the provider answers with `400 invalid parameters` are now reshaped on the way out instead of failing the whole turn.
+
+  - **A model narrates a tool call instead of making one.** The reply is recorded as pre-tool commentary, and every later message replayed that turn back to the provider was rejected — the session could not continue or be resumed, because each retry sent the same history again.
+  - **A host asks for a specific tool or forces a tool call.** The free provider accepts only automatic tool selection, so an unsupported choice is dropped rather than sent. Nothing is imposed on a host that did not ask.
+  - **A tool name carries a character the provider rejects.** Spaces, colons and slashes are folded for the trip out and the caller's own name is restored on the way back, so the tool still runs under the name it was registered with. Names that are already accepted are sent exactly as they always were.
+
+  Measured on Muse Spark 1.3: the same three requests that each failed with an error now answer normally, and a well-formed tool set reaches the model byte-identical to before.
+
+- 9085408: Three free models that no longer answer are gone from the picker
+
+  `cline-free/deepseek-v4.1-flash`, `cline-free/kimi-k3` and `cline-free/solar-pro4` were retired upstream, so every turn on them failed. All three were probed twice against the live service on 2026-10-13 and each time answered `404 model not found` ("Cline has no such model. Refresh the model list and retry."), while the two Cline models that survive — `cline-free/muse-spark-1.3-contributor` and `z-ai/glm-5.3-flash` — answered 200 with the same login and request shape. The login, the quota rollover and every other model are unchanged; the picker is now 36 models (10 OpenCode, 15 Kilo, 2 Cline, 9 OVH).
+
 ## 1.34.0
 
 ### Minor Changes

package/README.md

@@ -1,6 +1,6 @@
 # pi-freeflow
 
-> 30 free models with up to 1M context. No API keys to manage. Add your own relays to spread requests across more IPs.
+> 37 free models with up to 1M context. No API keys to manage. Add your own relays to spread requests across more IPs.
 
 Thin by design: a model list, a relay proxy, and a log. The host (`pi-ai`) handles thinking, normalization, and provider behavior.
 
@@ -16,10 +16,10 @@ Thin by design: a model list, a relay proxy, and a log. The host (`pi-ai`) handl
 
 | Feature | What it does | Cost |
 | :--- | :--- | :--- |
-| **30 free models** | 10 from OpenCode Zen, 15 from KiloCode Gateway, 5 from Cline, context windows up to 1M. Full list below. | **$0** |
-| **Relay pool** | Route requests through your own Cloudflare Workers and Vercel Edge relays. Traffic spreads across the pool, or stays on one sticky relay. A relay that rate-limits, times out, or drops the connection cools down while healthy ones take its traffic. | **$0** beyond your platforms' free tiers |
+| **37 free models** | 11 from OpenCode Zen, 15 from KiloCode Gateway, 2 from Cline, 9 from OVHcloud AI Endpoints, context windows up to 1M. Full list below. | **$0** |
+| **Relay pool** | Route requests through your own Cloudflare Workers, Deno Deploy, and Vercel Edge relays. Traffic spreads across the pool, or stays on one sticky relay. A relay that rate-limits, times out, or drops the connection cools down while healthy ones take its traffic. | **$0** beyond your platforms' free tiers |
 | **Automatic fallback** | When every relay is cooling down, requests go direct to upstream instead of failing. | **$0** |
-| **Short model names** | Every model has a slash-free, colon-free alias, plus an optional `:effort` suffix for thinking depth. You type `freeflow/<name>`. | **$0** |
+| **Short model names** | Most models have a slash-free, colon-free alias, plus an optional `:effort` suffix for thinking depth. You type `freeflow/<name>`; the two Cline models have no alias and must be typed in full. | **$0** |
 | **Shared local proxy** | One daemon on `127.0.0.1:28180` serves every session on the machine, so parallel subagents reuse it instead of opening their own connections. | **$0** |
 | **Logs you can read** | `~/.pi/agent/pi-freeflow.log`, rotated at 10MB. Tail it with `/freeflow logs`. | **$0** |
 
@@ -85,7 +85,7 @@ Manual fallback per platform:
 **Option A: Cloudflare Workers, auto deploy (recommended)**
 ```bash
 /freeflow deploy cloudflare  # prompts token in-memory, auto-adds to pool
-# or shorthand: /freeflow deploy (defaults to Cloudflare)
+# or /freeflow deploy with no argument to pick the platform interactively
 ```
 Manual fallback: `dash.cloudflare.com` → Workers → Create → Deploy → Edit code → paste the canonical worker source (see below) → Deploy → `/freeflow add https://your.workers.dev cf-worker-1`
 
@@ -110,7 +110,7 @@ For `api/relay.js`, use the canonical worker source (see below); `vercel.json` s
 
 **Canonical worker source (all platforms)**
 
-The relay worker template is generated per deployment by `/freeflow deploy`, with one shared core adapted for Vercel, Cloudflare, and Deno. Deployed relays accept requests from any pi-freeflow user by default, while only forwarding to an allowlist (`https://opencode.ai`, `https://api.kilo.ai`) with private-host and path checks.
+The relay worker template is generated per deployment by `/freeflow deploy`, with one shared core adapted for Vercel, Cloudflare, and Deno. Every deploy embeds a random 32-byte secret, and the generated worker answers `401` to any request without a matching `x-relay-auth` header; the client sends that secret to each relay in the pool. Only a hand-pasted legacy worker built without a secret is open to anyone. Either way, a relay forwards only to an allowlist (`https://opencode.ai`, `https://api.kilo.ai`) with private-host and path checks.
 
 ```js
 // Minimal Cloudflare illustration. Prefer /freeflow deploy: the generated
@@ -185,18 +185,19 @@ The same command set works identically in OMP and Pi:
 
 ---
 
-### 30 models, one command
+### 37 models, one command
 
 ```bash
 /model → freeflow → pick
 ```
 
-#### OpenCode Zen (10 models), Responses and Chat API
+#### OpenCode Zen (11 models), Responses and Chat API
 
 Good defaults for long coding sessions and agentic work.
 
 | Model ID | Creator / Lab | Context | Max Output | Thinking | Vision |
 | :--- | :--- | :--- | :--- | :--- | :--- |
+| `cat-jam-free` | Cat Jam | **256K** (262.144) | **131K** (131.072) | `low … xhigh` | ❌ |
 | `muse-spark-1.2-contributor-free` | Meta Superintelligence Labs | **1M** (1.048.576) | **131K** (131.072) | `minimal … xhigh` | ✅ |
 | `muse-spark-1.3-contributor-free` | Meta Superintelligence Labs | **1M** (1.048.576) | **131K** (131.072) | `minimal … xhigh` | ✅ |
 | `mimo-v2.6-flash-free` | Xiaomi MiMo | **1M** (1.048.576) | **131K** (131.072) | `minimal … xhigh`\* | ✅ |
@@ -208,7 +209,7 @@ Good defaults for long coding sessions and agentic work.
 | `step-5-preview-free` | StepFun | **1M** (1.000.000) | **64K** (65.536) | `low … high` | ✅ |
 | `ling-3.1-flash-free` | Inclusion AI | **262K** (262.144) | **32K** (32.768) | `minimal … max` | ❌ |
 
-#### KiloCode Gateway (16 models), OpenRouter compatible
+#### KiloCode Gateway (15 models), OpenRouter compatible
 
 Keyless access. Short aliases work for every row (the full ID is in parentheses).
 
@@ -226,13 +227,13 @@ Keyless access. Short aliases work for every row (the full ID is in parentheses)
 | `kilo-auto` (`kilo-auto/free`) | Kilo Gateway Auto | **256K** (256.000) | **10K** (10.000) | `minimal…xhigh`\* | ❌ |
 | `openrouter` (`openrouter/free`) | OpenRouter Free | **200K** (200.000) | **65K** (65.536) | `minimal…xhigh`\* | ✅ |
 | `glyph-cluster` (`stealth/glyph-cluster`) | Stealth preview (lab undisclosed) | **256K** (256.000) | **256K** (256.000) | `minimal…xhigh`\* | ❌ |
-| `step-5-preview:kilo` (`stepfun/...:free`) | StepFun | **1M** (1.000.000) | **64K** (64.000) | `minimal…xhigh`\* | ✅ |
+| `step-5-preview:kilo` (`stepfun/...:free`) | StepFun | **1M** (1.000.000) | **64K** (65.536) | `minimal…xhigh`\* | ✅ |
 | `inkling-small` (`thinkingmachines/inkling-small:free`) | Thinking Machines | **1M** (1.048.576) | **262K** (262.144) | `minimal…xhigh`\* | ✅ |
-| `ling-3.1-flash-kilo` (`inclusionai/ling-3.1-flash`) | Inclusion AI | **262K** (262.144) | **32K** (32.768) | `minimal…xhigh`\* | ❌ |
+| `ling-3.1-flash:kilo` (`inclusionai/ling-3.1-flash`) | Inclusion AI | **262K** (262.144) | **32K** (32.768) | `minimal…xhigh`\* | ❌ |
 
 \* Levels are forwarded as-is through the OpenRouter-style nested `reasoning` parameter; effort mapping is decided by each model. MiMo collapses `minimal→low` and `xhigh→high` upstream, so its selector shows 5 labels but only 3 distinct effort values.
 
-#### Cline (5 models), OpenAI compatible
+#### Cline (2 models), OpenAI compatible
 
 Rotating per-account promo, direct only (never through the relay pool). Sign in through your browser with `/freeflow cline login`, manage with `/freeflow cline accounts` and `/freeflow cline logout`. Requests roll across saved logins when one hits its daily free limit.
 
@@ -242,11 +243,28 @@ Each account can only be saved once: an extra login for an account you already h
 
 | Model ID | Creator / Lab | Context | Max Output | Thinking | Vision |
 | :--- | :--- | :--- | :--- | :--- | :--- |
-| `cline-free/deepseek-v4.1-flash` | DeepSeek | **1M** (1.000.000) | **384K** (384.000) | `low/high/max` | ✅ |
 | `cline-free/muse-spark-1.3-contributor` | Meta Superintelligence Labs | **1M** (1.048.576) | **131K** (131.072) | `minimal…xhigh`\* | ✅ |
-| `z-ai/glm-5.3-flash` | Zhipu AI | **1M** (1.000.000) | **131K** (131.072) | `low/high/max` | ✅ |
-| `cline-free/kimi-k3` | Moonshot AI | **1M** (1.048.576) | **131K** (131.072) | `low/high/max` | ✅ |
-| `cline-free/solar-pro4` | Upstage | **512K** (524.288) | **131K** (131.072) | `minimal…max`\* | ❌ |
+| `z-ai/glm-5.3-flash` | Zhipu AI | **1M** (1.048.576) | **131K** (131.072) | `low/high/max` | ✅ |
+
+#### OVHcloud AI Endpoints (9 models), OpenAI compatible
+
+Anonymous access — no account, no key, nothing to sign in to. Direct only (never through the relay pool), because OVH has no per-IP gate for relays to route around. OVH rate-limits anonymous callers per IP, so a burst of parallel subagents can hit `429 API rate limit exceeded`; the model list itself never costs anything.
+
+Pick them with the slash-free short name (the full ID is in parentheses). Only `low` and `high` thinking levels are offered, because those are the two OVH confirmed.
+
+All nine talk the normal chat API. Hosts that prefer the responses API work too — the proxy maps that shape to chat and maps the answer back, so tool calls and streamed output behave exactly as they do on the other providers.
+
+| Model ID | Creator / Lab | Context | Max Output | Thinking | Vision |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+| `gpt-oss-120b` (`ovh-free/gpt-oss-120b`) | OpenAI | **128K** (131.072) | **32K** (32.768) | `low/high` | ❌ |
+| `gpt-oss-20b` (`ovh-free/gpt-oss-20b`) | OpenAI | **128K** (131.072) | **32K** (32.768) | `low/high` | ❌ |
+| `qwen3.5-397b` (`ovh-free/Qwen3.5-397B-A17B`) | Alibaba Qwen | **256K** (262.144) | **64K** (65.536) | `low/high` | ✅ |
+| `qwen3.5-9b` (`ovh-free/Qwen3.5-9B`) | Alibaba Qwen | **256K** (262.144) | **64K** (65.536) | `low/high` | ✅ |
+| `qwen3.6-27b` (`ovh-free/Qwen3.6-27B`) | Alibaba Qwen | **256K** (262.144) | **64K** (65.536) | `low/high` | ✅ |
+| `qwen3.8-27b` (`ovh-free/Qwen3.8-27B`) | Alibaba Qwen | **256K** (262.144) | **32K** (32.768) | `low/high` | ✅ |
+| `mistral-small-3.2` (`ovh-free/Mistral-Small-3.2-24B-Instruct-2506`) | Mistral AI | **128K** (131.072) | **16K** (16.384) | ❌ | ✅ |
+| `mistral-7b-v0.3` (`ovh-free/Mistral-7B-Instruct-v0.3`) | Mistral AI | **64K** (65.536) | **64K** (65.536) | ❌ | ❌ |
+| `llama-3.3-70b` (`ovh-free/Meta-Llama-3_3-70B-Instruct`) | Meta | **128K** (131.072) | **16K** (16.384) | ❌ | ❌ |
 
 ---
 
@@ -264,16 +282,16 @@ cat ~/.pi/agent/pi-freeflow.log | tail -n 50
 
 ### Design
 
-This package stays thin. It ships three things: a model catalog, a relay proxy, and a log. There is no build step and zero runtime dependencies. Thinking and prompt normalization stay with the host (`pi-ai`).
+This package stays thin. It ships three things: a model catalog, a relay proxy, and a log. There is no build step and one runtime dependency (`undici`). Thinking and prompt normalization stay with the host (`pi-ai`).
 
-About 19k lines including tests. The full suite (sandboxed, network-mocked) and typecheck pass before every release. See CHANGELOG.md.
+About 43k lines including tests: `src/` is roughly 17k across 27 files, and the 78 test files add roughly 26k more. The full suite (sandboxed, network-mocked) and typecheck pass before every release. See CHANGELOG.md.
 
 ---
 
 ### FAQ
 
 **Do I need API keys?**
-No. Kilo uses a shared free credential and OpenCode free models need no header. You never paste a key. Cline free models need a browser login instead: `/freeflow cline login` saves it to a per-user pool on your machine.
+No. OpenCode and Kilo free models are both keyless: the proxy sends no `Authorization` header at all, and a placeholder credential is rejected rather than merely unused — `Bearer kilo-free` answers `401 INVALID_TOKEN`, while sending nothing answers 200 for every free catalog model. You never paste a key. Cline free models need a browser login instead: `/freeflow cline login` saves it to a per-user pool on your machine. OVHcloud AI Endpoints models need no account and no key at all.
 
 **What if all relays hit rate limits?**
 The proxy tries direct upstream. If that is also rate-limited, the host shows the limit. That number is the shared upstream cap; without relays you would hit the same wall sooner.
@@ -340,27 +358,42 @@ pnpm smoke       # verifies extensions/index.ts loads without crashing
 
 ```
 src/
-├── index.ts          # extension entry, lifecycle hooks
-├── models.ts         # 30-model catalog definitions
-├── catalog.ts        # model catalog cache (24h disk)
-├── proxy.ts          # local proxy server (127.0.0.1:28180)
-├── relay.ts          # relay fetch, rolling on retryable failures
-├── relay-state.ts    # relay pool state, health tracking
-├── stream-pipe.ts    # SSE stream piping and truncation resilience
-├── commands.ts       # /freeflow CLI subcommands
-├── deploy.ts         # guided relay deploy (vercel/cloudflare/deno)
-├── config.ts         # constants, whitelists, paths, and runtime settings
-├── logger.ts         # file logger with 10MB rotation
-└── types.ts          # shared type definitions
+├── index.ts                 # extension entry, lifecycle hooks
+├── models.ts                # 37-model catalog definitions
+├── catalog.ts               # model catalog cache (24h disk)
+├── proxy.ts                 # local proxy server (127.0.0.1:28180)
+├── client.ts                # session-side daemon attach/spawn and heartbeat
+├── daemon.ts                # detached proxy daemon process (owns the port)
+├── lease.ts                 # client lease registry and TTL retirement
+├── health.ts                # loopback /_health relay snapshot
+├── responses.ts             # Responses-API reasoning portability
+├── tool-translation.ts      # tool reshape across chat/responses/messages
+├── tool-args.ts             # schema-driven repair of model tool-call arguments
+├── relay.ts                 # relay fetch, rolling on retryable failures
+├── relay-state.ts           # relay pool state, health tracking
+├── probe.ts                 # relay reachability and latency probe
+├── upstream-health.ts       # Zen free-tier gating watch and Kilo failover
+├── stream-pipe.ts           # SSE stream piping and truncation resilience
+├── opencode-fingerprint.ts  # Zen free-tier client fingerprint enforcement
+├── system-ca-fetch.ts       # OS-trust-store retry for intercepted TLS
+├── commands.ts              # /freeflow CLI subcommands
+├── deploy.ts                # guided relay deploy (vercel/cloudflare/deno)
+├── cline-accounts.ts        # per-slot Cline key pool with failover
+├── cline-device-auth.ts     # Cline WorkOS device-flow login
+├── update-checker.ts        # npm registry version check, cached
+├── updater.ts               # package-manager choice for /freeflow update
+├── config.ts                # constants, whitelists, paths, and runtime settings
+├── logger.ts                # file logger with 10MB rotation
+└── types.ts                 # shared type definitions
 extensions/
-└── index.ts          # OMP/Pi extension manifest
+└── index.ts                 # OMP/Pi extension manifest
 test/
-└── *.test.ts         # mirrors src/, node:test runner
+└── *.test.ts                # mirrors src/, node:test runner
 ```
 
 #### Guidelines
 
-- **Stay thin.** Zero runtime dependencies, no build step. If it belongs in the host (`pi-ai`), don't add it here.
+- **Stay thin.** One runtime dependency (`undici`), no build step. If it belongs in the host (`pi-ai`), don't add it here.
 - **Test what you touch.** Every `src/*.ts` has a matching `test/*.test.ts`. Add or update tests for your change.
 - **Keep model IDs clean.** Slash-free, colon-free aliases for CLI compatibility. See existing patterns in `models.ts`.
 - **One concern per PR.** Bug fix? One PR. New relay platform? Separate PR. Easier to review, faster to merge.

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "pi-freeflow",
   "type": "module",
-  "version": "1.34.0",
+  "version": "1.36.0",
   "description": "Thin provider for OMP/Pi — model list + dumb relay proxy + log; host pi-ai owns thinking/normalization",
   "main": "extensions/index.ts",
   "types": "src/index.ts",

package/src/catalog.ts

@@ -66,6 +66,15 @@ import type {
  *   and responses from LLMs", not a chat model. Same exclusion class as
  *   jev-1.13-free: live, wrong shape for an agentic harness, so it must never
  *   re-enter the picker via disk cache or upstream merge.
+ * - cline-free/deepseek-v4.1-flash, cline-free/kimi-k3, cline-free/solar-pro4:
+ *   retired upstream (2026-10-13) — probed live twice each against the real
+ *   Cline pool, every attempt answered HTTP 404 `{"error":"model not found",
+ *   "success":false,"hint":"Cline has no such model. Refresh the model list and
+ *   retry."}`, while the two Cline models that survive (cline-free/
+ *   muse-spark-1.3-contributor and z-ai/glm-5.3-flash) answered 200 with the
+ *   same account and request shape. The lane is healthy; these three ids are
+ *   gone, so they must never re-enter the picker from disk cache or an
+ *   upstream merge.
  */
 export const DEAD_MODEL_IDS = new Set<string>([
  "jev-1.13-free",
@@ -93,6 +102,9 @@ export const DEAD_MODEL_IDS = new Set<string>([
  "fledge-alpha-free",
  "inclusionai/ling-3.0-flash-sante:free",
  "stepfun/step-3.7-flash:free",
+ "cline-free/deepseek-v4.1-flash",
+ "cline-free/kimi-k3",
+ "cline-free/solar-pro4",
 ]);
 /**
  * Free-tier allowlist for anything entering the picker via network or stale disk.
@@ -132,7 +144,7 @@ export function sanitizeCatalogModels(models: RegisteredModel[]): RegisteredMode
 }
 /**
  * In-memory cache of currently active/available free models.
- * Initialized with all 30 verified models for 0ms instant availability.
+ * Initialized with all 36 verified models for 0ms instant availability.
  */
 let aliveCatalog: RegisteredModel[] = ALL_MODELS.map((m) => ({
  ...m,

package/src/cline-accounts.ts

@@ -116,15 +116,17 @@ export const CLINE_CLIENT_VERSION = "3.5.54";
  * mirror reference/cline providers/request-headers.ts
  * (DEFAULT_CLINE_REQUEST_HEADERS) — no credential is derived from them.
  *
- * `X-CLIENT-TYPE` is the Cline desktop app's identity, and it is what the
- * free-model feed keys on (measured live 2026-09-22). The recommended-models
- * endpoint returns five free entries for `cline-sdk` / `cline-cli` / the
- * `VSCode Extension` label, but six for `cline-desktop` — the extra one being
- * `cline-free/kimi-k3`. The identity header alone flips that list
- * (`X-PLATFORM: Cline Desktop` does not; `X-CLIENT-TYPE` decides), and the
- * sixth model serves live chat completions (HTTP 200 + SSE) for the saved
- * per-user login under either identity. Version and User-Agent stay as-is:
- * the feed serves the six-model list with them unchanged.
+ * `X-CLIENT-TYPE` is the Cline desktop app's identity, and it gates the
+ * `cline-free/*` id space (measured live 2026-09-22: that prefix answers 403
+ * "only available via Cline product surfaces" with a bare bearer, and 200 with
+ * this set; every non-prefixed id answers 200 either way). The header is kept
+ * because one catalog id still depends on it — after the 2026-10-13 prune,
+ * `cline-free/muse-spark-1.3-contributor` is the last prefixed model in the
+ * catalog. It also decided the free-model feed at the time: five entries for
+ * `cline-sdk` / `cline-cli` / the `VSCode Extension` label against six for
+ * `cline-desktop`, the extra one being `cline-free/kimi-k3`. That sixth model
+ * is retired upstream and no longer requested. Version and User-Agent stay
+ * as-is: the feed served the desktop list with them unchanged.
  */
 export const CLINE_CLIENT_HEADERS: Record<string, string> = {
  "HTTP-Referer": "https://cline.bot",
@@ -138,12 +140,12 @@ export const CLINE_CLIENT_HEADERS: Record<string, string> = {
 /**
  * Free-limit report carried by a Cline 429 body, e.g.
  * {"error":{"code":"INFERENCE_CAP_ERROR","message":"Error 429: Daily free limit
- * reached on model deepseek/deepseek-v4.1-flash. Try again in 20h 4m"}}.
+ * reached on model meta/muse-spark-1.3-contributor. Try again in 20h 4m"}}.
  *
  * `modelId` is the name Cline spells in its own body, which is a different
- * namespace from the id this proxy requests (measured live 2026-09-22: we ask
- * for `cline-free/deepseek-v4.1-flash`, Cline answers about
- * `deepseek/deepseek-v4.1-flash`). The limit cache is therefore keyed by the
+ * namespace from the id this proxy requests (measured live 2026-10-13: we ask
+ * for `cline-free/muse-spark-1.3-contributor`, Cline answers about
+ * `meta/muse-spark-1.3-contributor`). The limit cache is therefore keyed by the
  * requested id; `modelId` is diagnostics only.
  */
 export interface ClineFreeLimit {

package/src/config.ts

@@ -31,6 +31,17 @@ export const OPENCODE_API_URL = `${UPSTREAM_OPENCODE}/v1`;
 // supplied at call time (never stored here).
 export const CLINE_API_BASE = "https://api.cline.bot/api/v1";
 export const CLINE_CHAT_URL = `${CLINE_API_BASE}/chat/completions`;
+// OVHcloud AI Endpoints serves OpenAI-compatible chat completions with no
+// credential at all (probed keyless 2026-10-13: no Authorization header,
+// 200). Direct-only: OVH has no Zen-style per-IP gate to route around, so
+// it never rides the relay pool and is deliberately absent from the deployed
+// relay allowlist (src/deploy.ts), which keeps the relay surface at exactly
+// two upstreams. Chat-only: OVH's responses endpoint rejects every array
+// input shape with 422 "input: data did not match any variant of untagged
+// enum ResponseInput" and accepts only a bare string (probed 2026-10-13), so
+// a responses-path caller is translated to chat upstream and back.
+export const OVH_API_BASE = "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1";
+export const OVH_CHAT_URL = `${OVH_API_BASE}/chat/completions`;
 
 // ── Network & Server defaults ───────────────────────────────────────
 export const DEFAULT_PORT = 28180;

package/src/index.ts

@@ -1,7 +1,7 @@
 /**
  * pi-freeflow — Modular, high-resiliency LLM extension for Pi & Oh My Pi (OMP)
  *
- * Provides access to 30 free models (10 OpenCode Zen + 15 KiloCode Gateway + 5 Cline) with:
+ * Provides access to 37 free models (11 OpenCode Zen + 15 KiloCode Gateway + 2 Cline + 9 OVH) with:
  * - Single-port daemon reuse on 28180 across concurrent subagents
  * - Multi-cloud rolling egress relays (Vercel Edge, Cloudflare, Deno)
  * - 0ms instant startup with verified static catalog and background live health checks
@@ -126,7 +126,7 @@ export function buildProviderConfig(
     id: m.id,
     name: m.name,
     api: m.api,
-    owned_by: m.source === "kilo" ? "kilocode" : m.source === "cline" ? "clinecode" : "opencode",
+    owned_by: m.source === "kilo" ? "kilocode" : m.source === "cline" ? "clinecode" : m.source === "ovh" ? "ovhcloud" : "opencode",
     reasoning: m.reasoning,
     thinking: m.reasoning
      ? {
@@ -146,7 +146,7 @@ export function buildProviderConfig(
      }
      : m.api === "openai-responses"
       ? { sessionAffinityFormat: "openai-nosession" }
-      : m.source === "kilo" || m.source === "cline"
+      : m.source === "kilo" || m.source === "cline" || m.source === "ovh"
        ? {
         supportsDeveloperRole: false,
         supportsReasoningEffort: !!m.thinkingLevelMap,