zevairouter 1.1.22 → 1.1.24

package/cli/app/CHANGELOG.md

@@ -1,5 +1,19 @@
 # Unreleased
 
+# v1.1.24 (2026-09-18)
+
+## Fixes
+- **Antigravity CLI no longer crashes mid-answer when a request is routed.** The router speaks OpenAI-style SSE, so it appends a `[DONE]` sentinel and may emit a bare `{"error":...}` chunk. Antigravity's language server dereferences `.response` on every chunk without a nil check, so either one segfaulted the CLI (`SIGSEGV` in `getStreamingTextCompletion`) right after the answer began streaming. MITM now forwards only chunks carrying a `response` object and ends the stream cleanly otherwise, which is how Google signals completion and failure on this endpoint. Forwarded chunks are passed through verbatim so opaque fields such as `thoughtSignature` are not perturbed.
+- **A cleanly finished Antigravity stream is no longer accounted as a failure.** Same-format passthrough only recognised the OpenAI terminal marker `choices[].finish_reason`, but a Cloud Code stream carries it as `response.candidates[].finishReason`. Successful answers were therefore recorded as `STREAM_MISSING_TERMINAL` (`FAILED 502`) and fed provider cooldown. Both shapes are now recognised, including a `promptFeedback.blockReason` terminal; a stream that genuinely ends without any terminal still fails.
+- **Reverted catalog injection for models the upstream account no longer offers.** v1.1.23 advertised every mapped model on `:fetchAvailableModels` so dropped tiers stayed selectable. Making an unavailable model selectable does not make it usable: the CLI sends the injected model's backend enum and Google answers `NOT_FOUND (404)` before any chat request is made, which is worse than the model being absent. The catalog is forwarded untouched again. Models the account cannot reach remain available through the router directly (for example `ag/gemini-3.8-flash-high`).
+
+# v1.1.23 (2026-09-18)
+
+## Fixes
+- **Mapped MITM models now stay selectable even after the upstream catalog drops them.** The CLI/IDE model picker is built from the upstream catalog response rather than from `mitmAlias`, so a model the router could serve became unselectable — and was reported as "no longer available" — once Google stopped offering it to the account. MITM now rewrites `:fetchAvailableModels` and `:listModelConfigs` to advertise every mapped model. Upstream stays authoritative: existing entries are never modified, absent aliases are cloned from an upstream entry so required fields keep upstream's shape, and any non-JSON, non-catalog, or error response is forwarded untouched.
+- **MITM model mappings are read from the same data directory the app writes them to.** `src/mitm/paths.js` resolved `DATA_DIR` under a stale product-named directory while the app writes `mitm/aliases.json` under `~/.zevai`, so `getMitmAlias()` always read `null` and every intercepted request silently fell through to passthrough. Left over from the earlier product renames, which only updated the app side.
+- **Antigravity Gemini 3.6 Flash requests route to their own tier.** The router exposes `ag/gemini-3.6-flash-{high,medium,low}`, but without matching alias slots the generic Flash patterns silently resolved 3.6 requests onto 3.5 slots, so the served model differed from the one the CLI reported.
+
 # v1.1.22 (2026-09-14)
 
 ## Fixes

package/cli/app/package.json

@@ -1,6 +1,6 @@
 {
   "name": "zevairouter",
-  "version": "1.1.22",
+  "version": "1.1.24",
   "description": "Self-hosted AI router: many providers, one OpenAI-compatible /v1 endpoint and dashboard",
   "bin": {
     "zevai": "cli/cli.js"