feat(discord): show per-tool glyphs on progress draft tool rows (#168032)

## What Problem This Solves

Discord progress drafts show tool-log rows without a per-tool glyph (`• Exec: running`), while Telegram shows `🛠️ Exec: running` since #167392. Discord already marks the 🧠 reasoning and 💬 commentary lanes, so its tool rows were the only unmarked lane.

## User Impact

Discord tool-log rows in progress drafts now start with the tool's text glyph (`🛠️ Exec`, `📖 Read`, `🩹 Apply Patch`). The 🧠 / 💬 lanes, plan rows, approval rows, and plain status rows are unchanged. No config changes. Slack and other channels keep plain rows.

## Why This Change Was Made

#167392 added the core opt-in (`toolIcons?: boolean` on `createChannelProgressDraftCompositor`, glyph table in `src/channels/progress-tool-glyphs.ts`) and left Discord as a follow-up. This PR sets `toolIcons: true` on every Discord compositor so all Discord paths render the same:

- live draft preview (`message-handler.draft-preview.ts`)
- retained progress continuation after handoff (`message-handler.progress-continuation.ts`)
- retained-card `edit` action rendering a progress snapshot (`runtime.messaging.messages.ts`)

No Discord-specific glyph code. Rows that already carry an explicit `icon` keep it.

## Tests

- `message-handler.draft-preview.rest.test.ts`: new regression through the real draft preview and Discord REST client: `🧠 _Planning the check_` / `💬 Looking at the workspace` / `🛠️ Exec: running` / `📖 Read: running`. Fails on main with `• Exec: running` / `• Read: running`.
- `message-bodies.test.ts`: a retained row's explicit icon (`🧪 Read`) wins over the tool glyph.
- `message-handler.process.draft-final.test.ts`: existing expectations updated to the glyph rows.

## Bounded cost

- One static `Record` lookup per core-built tool row on Discord (core opt-in from #167392).
- No new config keys, protocol fields, network calls, or extra Discord sends/edits; the draft text gains one glyph and a space per tool row, inside the existing per-message length bounds.
- Production: +3 lines. Tests: +65/−5. Docs: +3/−3. No extra Discord sends or edits in the runs above beyond the existing progress-draft cadence.

## Evidence

Isolated gateway (`HOME`/`OPENCLAW_STATE_DIR`/`OPENCLAW_CONFIG_PATH` under `/tmp`) with the real bundled Discord plugin against QA Lab's Crabline Discord provider server, `streaming.mode: "progress"`, `progress.toolProgress: true`. Rows below are the Discord REST bodies the plugin sent, from the Crabline provider recorder (`artifacts/crabline/discord-provider-server.jsonl`).

**Live model `openai/gpt-6-astra`** (`qa suite --provider-mode live-frontier --model openai/gpt-6-astra --channel-driver crabline --channel discord`), prompt asked for one exec then one read:

```
POST  "Working\n\n🛠️ Exec: running\n🛠️ Exec: running"
PATCH "Working\n\n🛠️ Exec: running\n🛠️ Exec"
PATCH "…\n\n🛠️ Exec\n🛠️ Exec\n🛠️ Exec: running"
PATCH "…\n\n🛠️ Exec\n🛠️ Exec\n🛠️ Exec\n📖 Read from glyph-proof.txt"
POST  final reply (separate message)
```

**Scripted model (`mock-openai`)**, commentary + one exec:

```
POST  "Working\n\n💬 SLACK-QA-COMMENTARY-200C8C6D"
PATCH "Working\n\n💬 SLACK-QA-COMMENTARY-200C8C6D\n🛠️ Exec: running"
PATCH "Working\n\n💬 SLACK-QA-COMMENTARY-200C8C6D\n🛠️ Exec"
POST  "SLACK-QA-COMMENTARY-DONE-200C8C6D"
```

The 💬 commentary lane keeps its prefix, and the final reply is one separate message, as before. Before this change, the same rows render as `• Exec: running` / `• Read: running` (the new REST regression fails on main with exactly that diff).

docs/channels/discord/messaging.md

@@ -99,7 +99,7 @@ Messages in the same channel continue reaching the [reply queue](/concepts/queue
     - `block` emits draft-sized chunks; tune size and breakpoints with `streaming.preview.chunk` (`minChars`, `maxChars`, `breakPreference`), clamped to `textChunkLimit`. An explicit non-`off` preview mode overrides inherited `agents.defaults.blockStreamingDefault: "on"`; explicit `streaming.block.enabled: true` overrides the preview. If a turn cannot use previews, inherited block delivery still applies.
     - `progress` keeps one editable status draft until final delivery. By default it is quiet: the agent's latest preamble or narration as a status headline, 💬 commentary and 🧠 reasoning when they stream, ✅ / ▸ / ▢ plan steps, and any approval request or failed command. Ordinary tool calls do not add rows.
     - Media, error, and explicit-reply finals cancel pending preview edits.
-    - `streaming.progress.toolProgress: true` adds the rolling tool log underneath the headline: rows such as `Bash: run tests` or `Web Search: for "query"` (default `false`). `streaming.preview.toolProgress` controls tool rows in `partial` and `block` modes, where they default to `true`.
+    - `streaming.progress.toolProgress: true` adds the rolling tool log underneath the headline: rows such as `🛠️ Exec: running` or `📖 Read from notes.txt` (default `false`). Each tool row starts with a text glyph for its tool. `streaming.preview.toolProgress` controls tool rows in `partial` and `block` modes, where they default to `true`.
     - `streaming.progress.commentary` (default `false`) opts into raw assistant commentary in the temporary progress draft. The default preamble/narration status line is independent of this option. Commentary is cleaned before display, stays transient, and does not change final answer delivery.
     - `streaming.progress.maxLineChars` controls the per-line progress preview budget. Prose is shortened on word boundaries; command and path details keep useful suffixes.
     - `streaming.preview.commandText` / `streaming.progress.commandText` controls command/exec detail in compact progress lines: `status` (default, tool label only) or `raw` (explicit command text).

docs/concepts/progress-drafts.md

@@ -26,8 +26,8 @@ requests. On Discord and Telegram, it also keeps the current operation and
 delegated task status visible without a detailed tool log or utility model.
 Intermediate tool failures and nonzero command exits stay out of the draft.
 Set `streaming.progress.toolProgress: true` to add a rolling tool log,
-including tool failures, with rows such as `Bash: run tests`. Telegram
-prefixes each tool row with a text glyph for its tool, such as
+including tool failures, with rows such as `Bash: run tests`. Discord and
+Telegram prefix each tool row with a text glyph for its tool, such as
 `📖 Read: from docs/index.md`; other channels keep plain rows. Channel plugins opt in
 through the `toolIcons` option of `createChannelProgressDraftCompositor`.
 

extensions/discord/src/actions/message-bodies.test.ts

@@ -278,6 +278,29 @@ describe("Discord retained progress edits", () => {
     expect(current.content.length).toBeLessThanOrEqual(100);
   });
 
+  it("keeps a retained row's explicit icon over the tool glyph", async () => {
+    await discordMessageActions.handleAction?.({
+      channel: "discord",
+      action: "edit",
+      cfg: {
+        channels: {
+          discord: {
+            token,
+            groupPolicy: "open",
+            streaming: { progress: { label: false, toolProgress: true } },
+          },
+        },
+      },
+      params: { to: `channel:${channelId}`, messageId, message: "Generic fallback" },
+      progressSnapshot: {
+        lines: [{ kind: "tool", text: "Read", label: "Read", toolName: "read", icon: "🧪" }],
+      },
+      conversationReadOrigin: "direct-operator",
+    });
+
+    expect(current.content).toBe("🧪 Read");
+  });
+
   it.each([true, false])(
     "uses the current DM channel only for its captured user target (matching: %s)",
     async (matching) => {

extensions/discord/src/actions/runtime.messaging.messages.ts

@@ -125,6 +125,7 @@ export async function handleDiscordMessageManagementAction(ctx: DiscordMessaging
               active: true,
               seed: `${ctx.accountId}:${channelId}`,
               initialSnapshot: snapshot,
+              toolIcons: true,
             }).getText(),
             {
               maxChars: Math.min(ctx.accountConfig.textChunkLimit ?? 2000, 2000),

extensions/discord/src/monitor/message-handler.draft-preview.rest.test.ts

@@ -213,6 +213,41 @@ describe("Discord draft preview REST lifecycle", () => {
     }
   });
 
+  it("prefixes tool rows with tool glyphs and keeps the reasoning and commentary lanes", async () => {
+    vi.useFakeTimers();
+    const { controller, visible } = createContinuationHarness({
+      discordConfig: {
+        streaming: {
+          mode: "progress",
+          progress: { label: false, toolProgress: true, commentary: true },
+        },
+      },
+    });
+    try {
+      await controller.pushReasoningProgress("Planning the check");
+      await controller.pushItemEvent({
+        itemId: "preamble-1",
+        kind: "preamble",
+        phase: "end",
+        progressText: "Looking at the workspace",
+      });
+      await controller.pushItemEvent(
+        projectAgentToolActivity({ toolCallId: "exec-1", name: "exec", phase: "start" }),
+      );
+      await controller.pushItemEvent(
+        projectAgentToolActivity({ toolCallId: "read-1", name: "read", phase: "start" }),
+      );
+      await vi.advanceTimersByTimeAsync(1_500);
+      await controller.flush();
+      expect([...visible.values()]).toEqual([
+        "🧠 _Planning the check_\n💬 Looking at the workspace\n🛠️ Exec: running\n📖 Read: running",
+      ]);
+    } finally {
+      await controller.cleanup();
+      vi.useRealTimers();
+    }
+  });
+
   it("keeps one confirmed card after yield and isolates a queued turn's cleanup", async () => {
     vi.useFakeTimers();
     const harness = createContinuationHarness({ discordConfig: {} });

extensions/discord/src/monitor/message-handler.draft-preview.ts

@@ -125,6 +125,7 @@ export function createDiscordDraftPreviewController(params: {
     seed: progressSeed,
     reasoningLinePrefix: "🧠 ",
     commentaryLinePrefix: "💬 ",
+    toolIcons: true,
     commentaryItalics: false,
     update: async (previewText, options) => {
       if (!draftStream) {

extensions/discord/src/monitor/message-handler.process.draft-final.test.ts

@@ -209,7 +209,7 @@ describe("processDiscordMessage draft streaming final delivery", () => {
 
     const updates = draftStream.update.mock.calls.map((call) => call[0]);
     expect(updates).toContain(
-      "Reading the gateway config and restarting agents.\n\n• Exec: running",
+      "Reading the gateway config and restarting agents.\n\n🛠️ Exec: running",
     );
     expectFinalAnswerText("done");
   });
@@ -281,7 +281,7 @@ describe("processDiscordMessage draft streaming final delivery", () => {
     await runProcessDiscordMessage(ctx);
 
     expect(getLastDispatchReplyOptions()?.sourceReplyDeliveryMode).toBe("message_tool_only");
-    expect(draftStream.update).toHaveBeenCalledWith("Working\n\n• Exec: running\n• exec done", {
+    expect(draftStream.update).toHaveBeenCalledWith("Working\n\n🛠️ Exec: running\n• exec done", {
       complete: true,
     });
     expect(deliverDiscordReply).not.toHaveBeenCalled();
@@ -611,7 +611,9 @@ describe("processDiscordMessage draft streaming progress", () => {
     await runProgressScenario({ toolProgress: true, label: "Investigating", commandText: "raw" });
 
     expect(draftStream.retarget).toHaveBeenCalledWith("thread-1");
-    expect(draftStream.update).toHaveBeenLastCalledWith("Investigating\n\n• Checked the pipeline.");
+    expect(draftStream.update).toHaveBeenLastCalledWith(
+      "Investigating\n\n🛠️ Checked the pipeline.",
+    );
     expect(draftStream.messageId()).toBeDefined();
     expect(deliverDiscordReply).not.toHaveBeenCalled();
   });
@@ -764,7 +766,7 @@ describe("processDiscordMessage draft streaming reasoning", () => {
 
     await runProgressScenario({ toolProgress: true, label: "Clawing..." });
 
-    expect(draftStream.update).toHaveBeenCalledExactlyOnceWith("Clawing...\n\n• Apply Patch", {
+    expect(draftStream.update).toHaveBeenCalledExactlyOnceWith("Clawing...\n\n🩹 Apply Patch", {
       complete: true,
     });
   });
@@ -786,7 +788,7 @@ describe("processDiscordMessage draft streaming reasoning", () => {
 
     await runProgressScenario({ toolProgress: true, label: "Clawing..." });
 
-    expect(draftStream.update).toHaveBeenCalledWith("Clawing...\n\n• Exec: running\n• done", {
+    expect(draftStream.update).toHaveBeenCalledWith("Clawing...\n\n🛠️ Exec: running\n• done", {
       complete: true,
     });
     expect(draftStream.update.mock.calls.map((call) => call[0]).join("\n")).not.toContain(

extensions/discord/src/monitor/message-handler.progress-continuation.ts

@@ -40,6 +40,7 @@ export function retainDiscordProgressDraft(params: {
     active: true,
     seed: params.seed,
     reasoningGate: false,
+    toolIcons: true,
     initialSnapshot: params.snapshot,
     update: async (text, options) => {
       if (retired) {