@dreb/coding-agent 2.61.2 → 2.63.0

package/docs/rpc.md

@@ -1325,7 +1325,7 @@ Note: with `summarize: true` the command is LLM-bound and can take a while. `Rpc
 
 Persistent settings, backed by the settings file (see [settings.md](settings.md)). They are normally distinct from live session state, with global-only control/security-policy exceptions:
 
-- **Persistent defaults** (`get_settings` / `set_settings`): provider/model, thinking level, queue modes, compaction/retry/image/skill/thinking-display/transport toggles, `maxConcurrentSubagents`, `enabledModels`, and per-agent model fallback lists seed fresh runtimes. Writing these ordinary defaults does **not** change a running session.
+- **Persistent defaults** (`get_settings` / `set_settings`): provider/model, thinking level, queue modes, compaction/retry/image/skill/thinking-display/transport toggles, `maxConcurrentSubagents`, `enabledModels`, `tabTitle`, and per-agent model fallback lists seed fresh runtimes. Writing these ordinary defaults does **not** change a running session.
 - **Global nested-context trust policy** (`autoLoadNestedContext`, `trustedContextFolders`, `effectiveTrustedContextRoots`, and the trust commands below): this is read from `~/.dreb/agent/settings.json` only, never project settings. Active main/subagent processes observe it for **future lazy nested/out-of-cwd loads**; it cannot remove content already injected into a conversation. It does not govern the separate initial upward context scan from the launch cwd.
 - **Global Dispatch Arbiter policy** (`subagentArbiter`): the complete object is read/written globally and project settings cannot shadow it. Enabled runtimes consume it before future subagent spawns; it does not rewrite already-started children.
 - **Runtime state** (`get_state` / `set_model` / `set_thinking_level` / `set_steering_mode` / `set_follow_up_mode` / `set_auto_compaction` / `set_auto_retry`): the state of the live session. Note that the runtime setters also persist their values as new defaults as a side effect.
@@ -1379,6 +1379,12 @@ Response:
       "model": "anthropic/claude-sonnet-4-5",
       "thinking": "medium",
       "guidePath": "~/.dreb/agent/model-routing-guide.md"
+    },
+    "tabTitle": {
+      "enabled": true,
+      "model": "anthropic/claude-haiku-4-5",
+      "triggerAfter": 9,
+      "maxTitleLength": 60
     }
   }
 }
@@ -1392,6 +1398,8 @@ Response:
 
 `subagentArbiter` is absent when unconfigured. It is always the global object; project `.dreb/settings.json` cannot enable, disable, or alter it.
 
+`tabTitle` is the effective merged object and is absent when unconfigured. Its optional `model` is one exact `provider/model`; when absent, title generation preserves the Explore-agent resolution route documented in [settings.md](settings.md#tab-title).
+
 #### set_settings
 
 Update persistent default settings. Takes a partial payload — only the supplied keys change. The whole payload is validated before anything is applied: on any invalid field, nothing changes and the response is an explicit error. Writes target the global settings file (same scope as every runtime setter).
@@ -1448,6 +1456,16 @@ Setting per-agent model fallback lists:
 
 For `agentModels`, a non-empty array writes the global fallback list for that agent. An empty array removes the global entry, so that agent uses its agent-definition default unless a project-level override exists.
 
+Update one or more tab-title fields. Nested updates are partial, so changing `enabled` or `model` preserves omitted `triggerAfter` and `maxTitleLength` values. A supplied model must be an available exact reference; model IDs may contain `/` after the provider prefix. `model: null` removes the pinned model, restoring Explore-agent routing:
+
+```json
+{"type":"set_settings","settings":{"tabTitle":{"enabled":true,"model":"anthropic/claude-haiku-4-5"}}}
+```
+
+```json
+{"type":"set_settings","settings":{"tabTitle":{"model":null}}}
+```
+
 Replace the complete global-only Dispatch Arbiter policy (exact model is validated; explicit thinking is capability-validated):
 
 ```json
@@ -1537,6 +1555,7 @@ Valid keys and values:
 | `agentModels` | Plain object mapping agent names to arrays of non-empty model id strings; empty arrays remove the global entry for that agent |
 | `enabledModels` | Non-empty ordered array of available exact `provider/model` references, or explicit `null` to remove the global filter and restore implicit all. Duplicate, glob, fuzzy, and thinking-suffix entries are rejected. |
 | `subagentArbiter` | Complete global-only object or `null`. Keys: `enabled` boolean, exact available `model`, optional valid/capability-supported `thinking`, non-empty `guidePath`. Enabling requires `model`. Unknown nested keys are rejected. |
+| `tabTitle` | Partial object. Keys: `enabled` boolean, exact available `model` (`null` removes the pinned model and restores Explore-agent routing), positive whole-number `triggerAfter`, positive whole-number `maxTitleLength`. Unknown nested keys and an empty object are rejected. |
 
 Errors are explicit `success: false` responses (nothing is applied on any of them):
 
@@ -1552,6 +1571,7 @@ Errors are explicit `success: false` responses (nothing is applied on any of the
 - Provider without model (or vice versa): `defaultProvider and defaultModel must be set together`
 - Unavailable model: `Model not found: provider/model-id`
 - Invalid arbiter policy: `Enabling subagentArbiter requires an exact provider/model`, `Arbiter model not found: ...`, or a nested-key/type/thinking capability error
+- Invalid tab-title update: `tabTitle.model must be an exact provider/model`, `Tab title model not found: ...`, or a nested-key/type/positive-number error
 - Corrupt settings file: `Cannot write settings: the global settings file failed to load (fix or remove the corrupt settings.json first)` — without this guard the write would silently no-op
 - Write failure (I/O error): `Failed to persist settings: ...`
 

package/docs/settings.md

@@ -146,15 +146,19 @@ Prompt behavior:
 | Setting | Type | Default | Description |
 |---------|------|---------|-------------|
 | `tabTitle.enabled` | boolean | `true` | Auto-generate terminal tab title from session task |
+| `tabTitle.model` | string | - | Exact `provider/model` for the title call; absent preserves Explore-agent routing |
 | `tabTitle.triggerAfter` | number | `9` | Number of tool calls before generating title |
 | `tabTitle.maxTitleLength` | number | `60` | Soft target length hint for generated titles (clamped to a hard cap of 300) |
 
 After the configured number of tool calls, dreb fires a single background LLM call to summarize the session's task into a terminal tab title, then sets it via OSC 0. The title is based primarily on your actual request and current-session actions, with branch/repo/cwd metadata used only for disambiguation. `maxTitleLength` is a soft target communicated to the model (default 60); titles may run a little longer for clarity and are hard-capped at 300 characters. The TUI truncates long titles visually while the dashboard shows the full name. Only fires once per session, and never overwrites an already-named (e.g. resumed) session. If the LLM call fails, the default title remains and the failure is surfaced (shown in interactive mode, logged to stderr in RPC mode).
 
+`tabTitle.model` pins the primary call to one available exact model. Model IDs may contain `/`, so the first path segment is the provider and the complete remainder is the model ID. When this setting is absent, resolution is unchanged: the Explore `agentModels` override, then Explore agent frontmatter, then the parent session model. If the selected model call fails and differs from the parent session model, dreb retries once with the parent. Dashboard Settings exposes `enabled` and the exact model picker, including clearing a pinned model back to the automatic route; `triggerAfter` and `maxTitleLength` remain configurable in JSON.
+
 ```json
 {
   "tabTitle": {
     "enabled": true,
+    "model": "anthropic/claude-haiku-4-5",
     "triggerAfter": 9,
     "maxTitleLength": 60
   }

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@dreb/coding-agent",
-	"version": "2.61.2",
+	"version": "2.63.0",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"type": "module",
 	"drebConfig": {