feat(codex-link): bridge 7 - external models work like subscription models

Measured with the unmodified Codex binary on 10 Oct 2026:
- real context windows per model (NanoGPT list, local table, combo = its
  smallest member, capped at the summarizer's window) so Codex compacts by
  itself when switching to a smaller model; an overflow guard trims
  oversized old parts and reports what still does not fit
- classic tools instead of code mode for external models: GLM 0/3 -> 3/3,
  Qwen 0/3 -> 2/3 on simple tool tasks, Claude/cx/Gemini unchanged
- phase (commentary / final_answer) on external assistant messages
- an answer cut by the client is carried to the next turn once
- images stay for vision-capable targets
- summary brake stops tight loops, not frequent compaction
- permanent gateway errors and tool-less targets fail at once with a
  readable message instead of five reconnects
- a lone "." (used to apply a model change in voice) is answered locally
- the bridge listens within seconds; the IDE waits 45 s for it

Release: AITURK IDE 1.0.0-beta.26.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

apps/desktop/assets/codex-bridge/README.md

@@ -26,3 +26,25 @@ Run: `python bridge.py --home <data dir> --hermes-home <agent home>`. The IDE wr
 `POST /aiturk/codex-baglama/stop` ends the process.
 
 Tests: `python -m unittest discover` in this directory.
+
+## Version 7 (IDE beta.26): external models behave like subscription models
+
+Measured on 10 Oct 2026 with the unmodified Codex binary (`codex exec`, same conversation, model switched per turn).
+
+| Area | Before | Now |
+| --- | --- | --- |
+| Context window | every added model declared 272 000 | `pencere.py`: NanoGPT's own list for `nano/*`, a local table for `cc`/`cx`/`ag`, a combo declares its smallest member, unknown = 128 000, nothing above 272 000 (the summarizer's window). Codex then compacts by itself when switching to a smaller model. |
+| Overflow | the target answered with an empty stream; Codex retried five times | `pencere.sigdir` trims the middle of oversized old parts in the outgoing copy; if it still does not fit Codex gets `context_length_exceeded`. A limit reported by the target is learned (`pencere-ogrenilen.json`). |
+| Tools | clones inherited `code_mode_only` + Responses Lite; GLM and Qwen failed 0/3 on simple tasks (empty answers, reasoning leaked into the answer) | `dis_arac_modu: "direct"` (default): classic tools. GLM 3/3, Qwen 2/3; Claude, `cx` and Gemini unchanged (6/6). `"kalip"` restores the old behaviour. |
+| Intermediate messages | external assistant messages had no `phase`; Codex treats those as final answers | `SseRewriter` stamps `commentary` / `final_answer`. Deltas are never delayed; only the item's `done` event waits for what follows. |
+| Interrupted answer | text written before a stop or an app exit was lost | `yarim.py` keeps it (memory only) when the *client* closes the stream and adds it as a note to the next user message of the same conversation, once. |
+| Images | removed from every old message for all external models | kept for vision-capable targets (user messages, newest first, 1 MB total). |
+| Summary brake | the 4th summary in 10 minutes failed the turn | three summaries in a row with no normal request in between, or eight in 10 minutes. |
+| Gateway errors | `event: error` (e.g. `upstream 402`) looked like a cut stream | permanent ones become `response.failed` with a readable message; short upstream bodies are logged. |
+| Start-up | the model list was fetched before listening; a slow gateway made the IDE give up after 10 s | window data comes from the cache at start and is refreshed in the background; the IDE waits 45 s. |
+
+Not changed: subscription models, their order and names, and everything on the voice path (`/realtime/*` passes through untouched).
+
+Settings in `bridge.json` (all optional): `dis_arac_modu`, `tasma_bekcisi`, `yarim_cevap`, `ozet_kopru.dongu_siniri`, `ozet_kopru.dongu_art_arda`.
+
+A lone `.` sent as the user's message (used to apply a model change during a voice session) is answered by the bridge itself with "<model> hazır." and never reaches a model (`nokta_karsila`, default on). Targets that reject tool calls are remembered (`aracsiz-modeller.json`) and marked "araçsız" in the picker.