From 20b4e2b8595649e00cc5202036421f05becf9928 Mon Sep 17 00:00:00 2001
From: Peter Steinberger <steipete@gmail.com>
Date: Sun, 11 Jan 2026 02:24:35 +0000
Subject: [PATCH] fix: stabilize live probes and docs

---
 docs/automation/webhook.md                    |  2 +-
 docs/broadcast-groups.md                      | 16 +++++-
 docs/cli/gateway.md                           |  1 +
 docs/cli/index.md                             | 57 +++++++++++++++++--
 docs/cli/message.md                           | 26 ++++++++-
 docs/cli/sandbox.md                           | 28 ++++-----
 docs/gateway/configuration.md                 |  4 +-
 docs/providers/index.md                       |  2 +
 docs/testing.md                               | 11 +++-
 docs/tools/agent-send.md                      |  2 +-
 docs/tools/index.md                           |  6 +-
 patches/@mariozechner__pi-ai@0.42.2.patch     | 54 ------------------
 src/agents/models.profiles.live.test.ts       | 19 +++++++
 .../openai-responses.reasoning-replay.test.ts | 10 ++--
 14 files changed, 149 insertions(+), 89 deletions(-)

diff --git a/docs/automation/webhook.md b/docs/automation/webhook.md
index eae5e4a48..b3ec0d65e 100644
--- a/docs/automation/webhook.md
+++ b/docs/automation/webhook.md
@@ -72,7 +72,7 @@ Payload:
 - `wakeMode` optional (`now` | `next-heartbeat`): Whether to trigger an immediate heartbeat (default `now`) or wait for the next periodic check.
 - `deliver` optional (boolean): If `true`, the agent's response will be sent to the messaging provider. Defaults to `true`. Responses that are only heartbeat acknowledgments are automatically skipped.
 - `provider` optional (string): The messaging service for delivery. One of: `last`, `whatsapp`, `telegram`, `discord`, `slack`, `signal`, `imessage`, `msteams`. Defaults to `last`.
-- `to` optional (string): The recipient identifier for the provider (e.g., phone number for WhatsApp/Signal, chat ID for Telegram, channel ID for Discord/Slack). Defaults to the last recipient in the main session.
+- `to` optional (string): The recipient identifier for the provider (e.g., phone number for WhatsApp/Signal, chat ID for Telegram, channel ID for Discord/Slack, conversation ID for MS Teams). Defaults to the last recipient in the main session.
 - `model` optional (string): Model override (e.g., `anthropic/claude-3-5-sonnet` or an alias). Must be in the allowed model list if restricted.
 - `thinking` optional (string): Thinking level override (e.g., `low`, `medium`, `high`).
 - `timeoutSeconds` optional (number): Maximum duration for the agent run in seconds.
diff --git a/docs/broadcast-groups.md b/docs/broadcast-groups.md
index 58833dbdc..d698c3b5e 100644
--- a/docs/broadcast-groups.md
+++ b/docs/broadcast-groups.md
@@ -1,3 +1,11 @@
+---
+summary: "Broadcast a WhatsApp message to multiple agents"
+read_when:
+  - Configuring broadcast groups
+  - Debugging multi-agent replies in WhatsApp
+status: experimental
+---
+
 # Broadcast Groups
 
 **Status:** Experimental  
@@ -5,7 +13,9 @@
 
 ## Overview
 
-Broadcast Groups enable multiple agents to process and respond to the same message simultaneously. This allows you to create specialized agent teams that work together in a single WhatsApp group, DM, or channel - all using one phone number.
+Broadcast Groups enable multiple agents to process and respond to the same message simultaneously. This allows you to create specialized agent teams that work together in a single WhatsApp group or DM — all using one phone number.
+
+Current scope: **WhatsApp only** (web provider).
 
 Broadcast groups are evaluated after provider allowlists and group activation rules. In WhatsApp groups, this means broadcasts happen when Clawdbot would normally reply (for example: on mention, depending on your group settings).
 
@@ -54,7 +64,9 @@ Agents:
 
 ### Basic Setup
 
-Add a top-level `broadcast` section (next to `bindings`):
+Add a top-level `broadcast` section (next to `bindings`). Keys are WhatsApp peer ids:
+- group chats: group JID (e.g. `120363403215116621@g.us`)
+- DMs: E.164 phone number (e.g. `+15551234567`)
 
 ```json
 {
diff --git a/docs/cli/gateway.md b/docs/cli/gateway.md
index e04c065d7..16e40ecf7 100644
--- a/docs/cli/gateway.md
+++ b/docs/cli/gateway.md
@@ -39,6 +39,7 @@ Notes:
 - `--password <password>`: password override (also sets `CLAWDBOT_GATEWAY_PASSWORD` for the process).
 - `--tailscale <off|serve|funnel>`: expose the Gateway via Tailscale.
 - `--tailscale-reset-on-exit`: reset Tailscale serve/funnel config on shutdown.
+- `--allow-unconfigured`: allow gateway start without `gateway.mode=local` in config.
 - `--dev`: create a dev config + workspace if missing (skips BOOTSTRAP.md).
 - `--reset`: reset dev config + credentials + sessions + workspace (requires `--dev`).
 - `--force`: kill any existing listener on the selected port before starting.
diff --git a/docs/cli/index.md b/docs/cli/index.md
index 9b0fb9201..571d1245c 100644
--- a/docs/cli/index.md
+++ b/docs/cli/index.md
@@ -52,6 +52,7 @@ clawdbot [--dev] [--profile <name>] <command>
   providers
     list
     status
+    logs
     add
     remove
     login
@@ -74,6 +75,14 @@ clawdbot [--dev] [--profile <name>] <command>
     health
     status
     discover
+  daemon
+    status
+    install
+    uninstall
+    start
+    stop
+    restart
+  logs
   models
     list
     status
@@ -83,6 +92,12 @@ clawdbot [--dev] [--profile <name>] <command>
     fallbacks list|add|remove|clear
     image-fallbacks list|add|remove|clear
     scan
+    auth add|setup-token|paste-token
+    auth order get|set|clear
+  sandbox
+    list
+    recreate
+    explain
   wake
   cron
     status
@@ -106,7 +121,8 @@ clawdbot [--dev] [--profile <name>] <command>
     run
     notify
     camera list|snap|clip
-    canvas snapshot
+    canvas snapshot|present|hide|navigate|eval
+    canvas a2ui push|reset
     screen record
     location get
   browser
@@ -180,6 +196,7 @@ Options:
 - `--workspace <dir>`
 - `--non-interactive`
 - `--mode <local|remote>`
+- `--flow <quickstart|advanced>`
 - `--auth-choice <setup-token|claude-cli|token|openai-codex|openai-api-key|codex-cli|antigravity|gemini-api-key|zai-api-key|apiKey|minimax-cloud|minimax-api|minimax|opencode-zen|skip>`
 - `--token-provider <id>` (non-interactive; used with `--auth-choice token`)
 - `--token <token>` (non-interactive; used with `--auth-choice token`)
@@ -201,9 +218,12 @@ Options:
 - `--tailscale <off|serve|funnel>`
 - `--tailscale-reset-on-exit`
 - `--install-daemon`
+- `--no-install-daemon` (alias: `--skip-daemon`)
 - `--daemon-runtime <node|bun>`
+- `--skip-providers`
 - `--skip-skills`
 - `--skip-health`
+- `--skip-ui`
 - `--node-manager <npm|pnpm|bun>`
 - `--json`
 
@@ -222,19 +242,20 @@ Options:
 ## Provider helpers
 
 ### `providers`
-Manage chat provider accounts (WhatsApp/Telegram/Discord/Slack/Signal/iMessage).
+Manage chat provider accounts (WhatsApp/Telegram/Discord/Slack/Signal/iMessage/MS Teams).
 
 Subcommands:
 - `providers list`: show configured chat providers and auth profiles (Claude Code + Codex CLI OAuth sync included).
 - `providers status`: check gateway reachability and provider health (`--probe` runs extra checks; use `clawdbot health` or `clawdbot status --deep` for gateway health probes).
 - Tip: `providers status` prints warnings with suggested fixes when it can detect common misconfigurations (then points you to `clawdbot doctor`).
+- `providers logs`: show recent provider logs from the gateway log file.
 - `providers add`: wizard-style setup when no flags are passed; flags switch to non-interactive mode.
 - `providers remove`: disable by default; pass `--delete` to remove config entries without prompts.
 - `providers login`: interactive provider login (WhatsApp Web only).
 - `providers logout`: log out of a provider session (WhatsApp Web only).
 
 Common options:
-- `--provider <name>`: `whatsapp|telegram|discord|slack|signal|imessage`
+- `--provider <name>`: `whatsapp|telegram|discord|slack|signal|imessage|msteams`
 - `--account <id>`: provider account id (default `default`)
 - `--name <label>`: display name for the account
 
@@ -251,6 +272,11 @@ Common options:
 - `--no-usage`: skip provider usage/quota snapshots (OAuth/API-backed only).
 - `--json`: output JSON (includes usage unless `--no-usage` is set).
 
+`providers logs` options:
+- `--provider <name|all>` (default `all`)
+- `--lines <n>` (default `200`)
+- `--json`
+
 OAuth sync sources:
 - Claude Code → `anthropic:claude-cli`
   - macOS: Keychain item "Claude Code-credentials" (choose "Always Allow" to avoid launchd prompts)
@@ -336,7 +362,7 @@ Options:
 - `--session-id <id>`
 - `--thinking <off|minimal|low|medium|high>`
 - `--verbose <on|off>`
-- `--provider <whatsapp|telegram|discord|slack|signal|imessage>`
+- `--provider <whatsapp|telegram|discord|slack|signal|imessage|msteams>`
 - `--local`
 - `--deliver`
 - `--json`
@@ -377,10 +403,12 @@ Show linked session health and recent recipients.
 
 Options:
 - `--json`
+- `--all` (full diagnosis; read-only, pasteable)
 - `--deep` (probe providers)
 - `--usage` (show provider usage/quota)
 - `--timeout <ms>`
 - `--verbose`
+- `--debug` (alias for `--verbose`)
 
 ### Usage tracking
 Clawdbot can surface provider usage/quota when OAuth/API creds are available.
@@ -431,8 +459,11 @@ Options:
 - `--reset` (reset dev config + credentials + sessions + workspace)
 - `--force` (kill existing listener on port)
 - `--verbose`
+- `--claude-cli-logs`
 - `--ws-log <auto|full|compact>`
 - `--compact` (alias for `--ws-log compact`)
+- `--raw-stream`
+- `--raw-stream-path <path>`
 
 ### `daemon`
 Manage the Gateway service (launchd/systemd/schtasks).
@@ -476,6 +507,7 @@ Subcommands:
 - `gateway call <method> [--params <json>]`
 - `gateway health`
 - `gateway status`
+- `gateway discover`
 
 Common RPCs:
 - `config.apply` (validate + write config + restart + wake)
@@ -495,6 +527,10 @@ clawdbot models status
 ### `models` (root)
 `clawdbot models` is an alias for `models status`.
 
+Root options:
+- `--status-json` (alias for `models status --json`)
+- `--status-plain` (alias for `models status --plain`)
+
 ### `models list`
 Options:
 - `--all`
@@ -545,12 +581,25 @@ Options:
 - `--max-candidates <n>`
 - `--timeout <ms>`
 - `--concurrency <n>`
+- `--no-probe`
 - `--yes`
 - `--no-input`
 - `--set-default`
 - `--set-image`
 - `--json`
 
+### `models auth add|setup-token|paste-token`
+Options:
+- `add`: interactive auth helper
+- `setup-token`: `--provider <name>` (default `anthropic`), `--yes`
+- `paste-token`: `--provider <name>`, `--profile-id <id>`, `--expires-in <duration>`
+
+### `models auth order get|set|clear`
+Options:
+- `get`: `--provider <name>`, `--agent <id>`, `--json`
+- `set`: `--provider <name>`, `--agent <id>`, `<profileIds...>`
+- `clear`: `--provider <name>`, `--agent <id>`
+
 ## Cron + wake
 
 ### `wake`
diff --git a/docs/cli/message.md b/docs/cli/message.md
index 44619d7e1..46b91f7cc 100644
--- a/docs/cli/message.md
+++ b/docs/cli/message.md
@@ -24,8 +24,8 @@ Provider selection:
 Target formats (`--to`):
 - WhatsApp: E.164 or group JID
 - Telegram: chat id or `@username`
-- Discord/Slack: `channel:<id>` or `user:<id>` (raw id is ambiguous for Discord)
-- Signal: E.164, `group:<id>`, or `signal:+E.164`
+- Discord/Slack: `channel:<id>` or `user:<id>` (raw `channelId` is also accepted)
+- Signal: `+E.164`, `group:<id>`, `signal:+E.164`, `signal:group:<id>`, or `username:<name>`/`u:<name>`
 - iMessage: handle or `chat_id:<id>`
 - MS Teams: conversation id (`19:...@thread.tacv2`) or `conversation:<id>` or `user:<aad-object-id>`
 
@@ -42,61 +42,79 @@ Target formats (`--to`):
 ### Core
 
 - `send`
+  - Providers: WhatsApp/Telegram/Discord/Slack/Signal/iMessage/MS Teams
   - Required: `--to`, `--message`
   - Optional: `--media`, `--reply-to`, `--thread-id`, `--gif-playback`
   - Telegram only: `--buttons-json` (requires `"inlineButtons"` in `telegram.capabilities` or `telegram.accounts.<id>.capabilities`)
+  - Telegram only: `--thread-id` (forum topic id)
+  - WhatsApp only: `--gif-playback`
 
 - `poll`
+  - Providers: WhatsApp/Discord/MS Teams
   - Required: `--to`, `--poll-question`, `--poll-option` (repeat)
   - Optional: `--poll-multi`, `--poll-duration-hours`, `--message`
+  - Discord only: `--poll-duration-hours`
 
 - `react`
+  - Providers: Discord/Slack/WhatsApp
   - Required: `--to`, `--message-id`
   - Optional: `--emoji`, `--remove`, `--participant`, `--from-me`, `--channel-id`
+  - WhatsApp only: `--participant`, `--from-me`
 
 - `reactions`
+  - Providers: Discord/Slack
   - Required: `--to`, `--message-id`
   - Optional: `--limit`, `--channel-id`
 
 - `read`
+  - Providers: Discord/Slack
   - Required: `--to`
   - Optional: `--limit`, `--before`, `--after`, `--around`, `--channel-id`
 
 - `edit`
+  - Providers: Discord/Slack
   - Required: `--to`, `--message-id`, `--message`
   - Optional: `--channel-id`
 
 - `delete`
+  - Providers: Discord/Slack
   - Required: `--to`, `--message-id`
   - Optional: `--channel-id`
 
 - `pin` / `unpin`
+  - Providers: Discord/Slack
   - Required: `--to`, `--message-id`
   - Optional: `--channel-id`
 
 - `pins` (list)
+  - Providers: Discord/Slack
   - Required: `--to`
   - Optional: `--channel-id`
 
 - `permissions`
+  - Providers: Discord
   - Required: `--to`
   - Optional: `--channel-id`
 
 - `search`
+  - Providers: Discord
   - Required: `--guild-id`, `--query`
   - Optional: `--channel-id`, `--channel-ids` (repeat), `--author-id`, `--author-ids` (repeat), `--limit`
 
 ### Threads
 
 - `thread create`
+  - Providers: Discord
   - Required: `--thread-name`, `--to` (channel id) or `--channel-id`
   - Optional: `--message-id`, `--auto-archive-min`
 
 - `thread list`
+  - Providers: Discord
   - Required: `--guild-id`
   - Optional: `--channel-id`, `--include-archived`, `--before`, `--limit`
 
 - `thread reply`
+  - Providers: Discord
   - Required: `--to` (thread id), `--message`
   - Optional: `--media`, `--reply-to`
 
@@ -104,18 +122,22 @@ Target formats (`--to`):
 
 - `emoji list`
   - Discord: `--guild-id`
+  - Slack: no extra flags
 
 - `emoji upload`
+  - Providers: Discord
   - Required: `--guild-id`, `--emoji-name`, `--media`
   - Optional: `--role-ids` (repeat)
 
 ### Stickers
 
 - `sticker send`
+  - Providers: Discord
   - Required: `--to`, `--sticker-id` (repeat)
   - Optional: `--message`
 
 - `sticker upload`
+  - Providers: Discord
   - Required: `--guild-id`, `--sticker-name`, `--sticker-desc`, `--sticker-tags`, `--media`
 
 ### Roles / Channels / Members / Voice
diff --git a/docs/cli/sandbox.md b/docs/cli/sandbox.md
index c167216ab..7e2a26adb 100644
--- a/docs/cli/sandbox.md
+++ b/docs/cli/sandbox.md
@@ -107,22 +107,24 @@ clawdbot sandbox recreate --agent alfred
 
 ## Configuration
 
-Sandbox settings are in `clawdbot.config.json`:
+Sandbox settings live in `~/.clawdbot/clawdbot.json` under `agents.defaults.sandbox` (per-agent overrides go in `agents.list[].sandbox`):
 
 ```jsonc
 {
-  "agent": {
-    "sandbox": {
-      "mode": "all",                    // off, non-main, all
-      "scope": "agent",                 // session, agent, shared
-      "docker": {
-        "image": "clawdbot-sandbox:bookworm-slim",
-        "containerPrefix": "clawdbot-sbx-"
-        // ... more Docker options
-      },
-      "prune": {
-        "idleHours": 24,               // Auto-prune after 24h idle
-        "maxAgeDays": 7                // Auto-prune after 7 days
+  "agents": {
+    "defaults": {
+      "sandbox": {
+        "mode": "all",                    // off, non-main, all
+        "scope": "agent",                 // session, agent, shared
+        "docker": {
+          "image": "clawdbot-sandbox:bookworm-slim",
+          "containerPrefix": "clawdbot-sbx-"
+          // ... more Docker options
+        },
+        "prune": {
+          "idleHours": 24,               // Auto-prune after 24h idle
+          "maxAgeDays": 7                // Auto-prune after 7 days
+        }
       }
     }
   }
diff --git a/docs/gateway/configuration.md b/docs/gateway/configuration.md
index cb4133359..7c76b1861 100644
--- a/docs/gateway/configuration.md
+++ b/docs/gateway/configuration.md
@@ -2018,7 +2018,7 @@ Mapping notes:
 - Templates like `{{messages[0].subject}}` read from the payload.
 - `transform` can point to a JS/TS module that returns a hook action.
 - `deliver: true` sends the final reply to a provider; `provider` defaults to `last` (falls back to WhatsApp).
-- If there is no prior delivery route, set `provider` + `to` explicitly (required for Telegram/Discord/Slack/Signal/iMessage).
+- If there is no prior delivery route, set `provider` + `to` explicitly (required for Telegram/Discord/Slack/Signal/iMessage/MS Teams).
 - `model` overrides the LLM for this hook run (`provider/model` or alias; must be allowed if `agents.defaults.models` is set).
 
 Gmail helper config (used by `clawdbot hooks gmail setup` / `run`):
@@ -2171,7 +2171,7 @@ Template placeholders are expanded in `tools.audio.transcription.args` (and any
 | `{{GroupMembers}}` | Group members preview (best effort) |
 | `{{SenderName}}` | Sender display name (best effort) |
 | `{{SenderE164}}` | Sender phone number (best effort) |
-| `{{Provider}}` | Provider hint (whatsapp|telegram|discord|imessage|webchat|…) |
+| `{{Provider}}` | Provider hint (whatsapp|telegram|discord|slack|signal|imessage|msteams|webchat|…) |
 
 ## Cron (Gateway scheduler)
 
diff --git a/docs/providers/index.md b/docs/providers/index.md
index 271853677..2e229015c 100644
--- a/docs/providers/index.md
+++ b/docs/providers/index.md
@@ -25,4 +25,6 @@ Text is supported everywhere; media and reactions vary by provider.
 - Providers can run simultaneously; configure multiple and Clawdbot will route per chat.
 - Group behavior varies by provider; see [Groups](/concepts/groups).
 - DM pairing and allowlists are enforced for safety; see [Security](/gateway/security).
+- Telegram internals: [grammY notes](/providers/grammy).
+- Troubleshooting: [Provider troubleshooting](/providers/troubleshooting).
 - Model providers are documented separately; see [Model Providers](/providers/models).
diff --git a/docs/testing.md b/docs/testing.md
index 4afcfd408..d49a690c5 100644
--- a/docs/testing.md
+++ b/docs/testing.md
@@ -133,7 +133,7 @@ Live tests are split into two layers so we can isolate failures:
 - Optional tool-calling stress:
   - `CLAWDBOT_LIVE_GATEWAY_TOOL_PROBE=1` enables an extra “bash writes file → read reads it back → echo nonce” check.
   - This is specifically meant to catch tool-calling compatibility issues across providers (formatting, history replay, tool_result pairing, etc.).
-  - Optional image send smoke:
+- Optional image send smoke:
   - `CLAWDBOT_LIVE_GATEWAY_IMAGE_PROBE=1` sends a real image attachment through the gateway agent pipeline (multimodal message) and asserts the model can read back a per-run code from the image.
   - Flow (high level):
     - Test generates a tiny PNG with “CAT” + random code (`src/gateway/live-image-probe.ts`)
@@ -142,6 +142,13 @@ Live tests are split into two layers so we can isolate failures:
     - Embedded agent forwards a multimodal user message to the model
     - Assertion: reply contains `cat` + the code (OCR tolerance: minor mistakes allowed)
 
+Tip: to see what you can test on your machine (and the exact `provider/model` ids), run:
+
+```bash
+pnpm clawdbot models list
+pnpm clawdbot models list --json
+```
+
 ## Live: Anthropic setup-token smoke
 
 - Test: `src/agents/anthropic.setup-token.live.test.ts`
@@ -225,7 +232,7 @@ This is the “common models” run we expect to keep working:
 - OpenAI (non-Codex): `openai/gpt-5.2` (optional: `openai/gpt-5.1`)
 - OpenAI Codex: `openai-codex/gpt-5.2` (optional: `openai-codex/gpt-5.2-codex`)
 - Anthropic: `anthropic/claude-opus-4-5` (or `anthropic/claude-sonnet-4-5`)
-- Google (Gemini API): `google/gemini-3-pro-preview` and `google/gemini-3-flash-preview`
+- Google (Gemini API): `google/gemini-3-pro-preview` and `google/gemini-3-flash-preview` (avoid older Gemini 2.x models)
 - Google (Antigravity): `google-antigravity/claude-opus-4-5-thinking` and `google-antigravity/gemini-3-flash`
 - Z.AI (GLM): `zai/glm-4.7`
 - MiniMax: `minimax/minimax-m2.1`
diff --git a/docs/tools/agent-send.md b/docs/tools/agent-send.md
index b01848ecb..3ec81a700 100644
--- a/docs/tools/agent-send.md
+++ b/docs/tools/agent-send.md
@@ -37,7 +37,7 @@ clawdbot agent --to +15555550123 --message "Summon reply" --deliver
 
 - `--local`: run locally (requires provider keys in your shell)
 - `--deliver`: send the reply to the chosen provider (requires `--to`)
-- `--provider`: `whatsapp|telegram|discord|slack|signal|imessage` (default: `whatsapp`)
+- `--provider`: `whatsapp|telegram|discord|slack|signal|imessage|msteams` (default: `whatsapp`)
 - `--thinking <off|minimal|low|medium|high>`: persist thinking level
 - `--verbose <on|off>`: persist verbose level
 - `--timeout <seconds>`: override agent timeout
diff --git a/docs/tools/index.md b/docs/tools/index.md
index a30d389c6..c7c2fb17c 100644
--- a/docs/tools/index.md
+++ b/docs/tools/index.md
@@ -145,11 +145,11 @@ Notes:
 - Uses the image model directly (independent of the main chat model).
 
 ### `message`
-Send messages and provider actions across Discord/Slack/Telegram/WhatsApp/Signal/iMessage.
+Send messages and provider actions across Discord/Slack/Telegram/WhatsApp/Signal/iMessage/MS Teams.
 
 Core actions:
 - `send` (text + optional media)
-- `poll` (WhatsApp/Discord polls)
+- `poll` (WhatsApp/Discord/MS Teams polls)
 - `react` / `reactions` / `read` / `edit` / `delete`
 - `pin` / `unpin` / `list-pins`
 - `permissions`
@@ -166,7 +166,7 @@ Core actions:
 
 Notes:
 - `send` routes WhatsApp via the Gateway; other providers go direct.
-- `poll` uses the Gateway for WhatsApp and direct Discord API for Discord.
+- `poll` uses the Gateway for WhatsApp and MS Teams; Discord polls go direct.
 
 ### `cron`
 Manage Gateway cron jobs and wakeups.
diff --git a/patches/@mariozechner__pi-ai@0.42.2.patch b/patches/@mariozechner__pi-ai@0.42.2.patch
index 956de72a0..0dee9b488 100644
--- a/patches/@mariozechner__pi-ai@0.42.2.patch
+++ b/patches/@mariozechner__pi-ai@0.42.2.patch
@@ -18,39 +18,6 @@ diff --git a/dist/providers/openai-codex-responses.js b/dist/providers/openai-co
 index 188a8294f26fe1bfe3fb298a7f58e4d8eaf2a529..a3aeb6a7ff53bc4f7f44362adb950b2c55455332 100644
 --- a/dist/providers/openai-codex-responses.js
 +++ b/dist/providers/openai-codex-responses.js
-@@ -433,9 +433,15 @@ function convertMessages(model, context) {
-         }
-         else if (msg.role === "assistant") {
-             const output = [];
-+            // OpenAI Responses rejects `reasoning` items that are not followed by a `message`.
-+            // Tool-call-only turns (thinking + function_call) are valid assistant turns, but
-+            // their stored reasoning items must not be replayed as standalone `reasoning` input.
-+            const hasTextBlock = msg.content.some((b) => b.type === "text");
-             for (const block of msg.content) {
-                 if (block.type === "thinking" && msg.stopReason !== "error") {
-                     if (block.thinkingSignature) {
-+                        if (!hasTextBlock)
-+                            continue;
-                         const reasoningItem = JSON.parse(block.thinkingSignature);
-                         output.push(reasoningItem);
-                     }
-@@ -470,6 +476,16 @@ function convertMessages(model, context) {
-             }
-             if (output.length === 0)
-                 continue;
-+            // OpenAI rejects standalone reasoning items when replaying a tool-only turn.
-+            // Only submit reasoning items when we also submit an assistant message item.
-+            // Repro: pnpm test src/agents/openai-responses.reasoning-replay.test.ts
-+            const hasMessage = output.some((item) => item?.type === "message");
-+            if (!hasMessage) {
-+                for (let i = output.length - 1; i >= 0; i -= 1) {
-+                    if (output[i]?.type === "reasoning")
-+                        output.splice(i, 1);
-+                }
-+            }
-             messages.push(...output);
-         }
-         else if (msg.role === "toolResult") {
 @@ -515,7 +531,7 @@ function convertTools(tools) {
          name: tool.name,
          description: tool.description,
@@ -60,24 +27,3 @@ index 188a8294f26fe1bfe3fb298a7f58e4d8eaf2a529..a3aeb6a7ff53bc4f7f44362adb950b2c
      }));
  }
  function mapStopReason(status) {
-diff --git a/dist/providers/openai-responses.js b/dist/providers/openai-responses.js
-index f07085c64390b211340d6a826b28ea9c2e77302f..523ed38a5a6151d6ff08dd89120315e7aaaf19b6 100644
---- a/dist/providers/openai-responses.js
-+++ b/dist/providers/openai-responses.js
-@@ -436,6 +436,16 @@ function convertMessages(model, context) {
-             }
-             if (output.length === 0)
-                 continue;
-+            // OpenAI rejects standalone reasoning items when replaying a tool-only turn.
-+            // Only submit reasoning items when we also submit an assistant message item.
-+            // Repro: pnpm test src/agents/openai-responses.reasoning-replay.test.ts
-+            const hasMessage = output.some((item) => item?.type === "message");
-+            if (!hasMessage) {
-+                for (let i = output.length - 1; i >= 0; i -= 1) {
-+                    if (output[i]?.type === "reasoning")
-+                        output.splice(i, 1);
-+                }
-+            }
-             messages.push(...output);
-         }
-         else if (msg.role === "toolResult") {
diff --git a/src/agents/models.profiles.live.test.ts b/src/agents/models.profiles.live.test.ts
index ee7b1c6c4..4ce3d7d54 100644
--- a/src/agents/models.profiles.live.test.ts
+++ b/src/agents/models.profiles.live.test.ts
@@ -48,6 +48,16 @@ function isGoogleModelNotFoundError(err: unknown): boolean {
   return false;
 }
 
+function isModelNotFoundErrorMessage(raw: string): boolean {
+  const msg = raw.trim();
+  if (!msg) return false;
+  if (/\b404\b/.test(msg) && /not[_-]?found/i.test(msg)) return true;
+  if (/not_found_error/i.test(msg)) return true;
+  if (/model:\s*[a-z0-9._-]+/i.test(msg) && /not[_-]?found/i.test(msg))
+    return true;
+  return false;
+}
+
 describeLive("live models (profile keys)", () => {
   it(
     "completes across configured models",
@@ -187,6 +197,15 @@ describeLive("live models (profile keys)", () => {
             },
           );
 
+          if (res.stopReason === "error") {
+            const msg = res.errorMessage ?? "";
+            if (ALL_MODELS && isModelNotFoundErrorMessage(msg)) {
+              skipped.push({ model: id, reason: msg });
+              continue;
+            }
+            throw new Error(msg || "model returned error with no message");
+          }
+
           const text = res.content
             .filter((block) => block.type === "text")
             .map((block) => block.text.trim())
diff --git a/src/agents/openai-responses.reasoning-replay.test.ts b/src/agents/openai-responses.reasoning-replay.test.ts
index 847438cd2..3ec326211 100644
--- a/src/agents/openai-responses.reasoning-replay.test.ts
+++ b/src/agents/openai-responses.reasoning-replay.test.ts
@@ -52,7 +52,7 @@ function installFailingFetchCapture() {
 }
 
 describe("openai-responses reasoning replay", () => {
-  it("handles tool-call-only turns without requiring reasoning replay", async () => {
+  it("replays reasoning for tool-call-only turns (required by OpenAI)", async () => {
     const cap = installFailingFetchCapture();
     try {
       const model = buildModel();
@@ -141,11 +141,11 @@ describe("openai-responses reasoning replay", () => {
         )
         .filter((t): t is string => typeof t === "string");
 
+      expect(types).toContain("reasoning");
       expect(types).toContain("function_call");
-      const reasoningIndex = types.indexOf("reasoning");
-      if (reasoningIndex !== -1) {
-        expect(reasoningIndex).toBeLessThan(types.indexOf("function_call"));
-      }
+      expect(types.indexOf("reasoning")).toBeLessThan(
+        types.indexOf("function_call"),
+      );
     } finally {
       cap.restore();
     }