revert: remove PR #4773 subagent restrictions feature

CONTEXT/PLAN-4709-restore-live-token-usage-2025-12-09.md

@@ -0,0 +1,209 @@
+# Plan: Restore Live Token Usage Feature (PR #4709)
+
+**Date:** 2025-12-09  
+**Related PR:** https://github.com/sst/opencode/pull/4709  
+**Status:** IMPLEMENTED - Feature restored 2025-12-10
+
+## Overview
+
+This plan documents the restoration of the "Live Token Usage During Streaming" feature that was originally added in PR #4709. The feature provides:
+
+- Real-time token tracking while streaming responses
+- `IN/OUT` format display for input/output tokens
+- Reasoning token display for "thinking" models
+- Toggle tokens command in TUI
+
+## Current State Analysis
+
+### What's Working
+
+| Component                      | File                                          | Status                     |
+| ------------------------------ | --------------------------------------------- | -------------------------- |
+| Token utility functions        | `packages/opencode/src/util/token.ts`         | **EXISTS**                 |
+| Message schema fields          | `packages/opencode/src/session/message-v2.ts` | **EXISTS**                 |
+| Token calculation in prompt.ts | `packages/opencode/src/session/prompt.ts`     | **EXISTS**                 |
+| Subtask completion fix         | `packages/opencode/src/session/prompt.ts`     | **EXISTS** (bug was fixed) |
+
+### What's Missing
+
+| Component                | File                                                         | Status      |
+| ------------------------ | ------------------------------------------------------------ | ----------- |
+| Streaming token updates  | `packages/opencode/src/session/processor.ts`                 | **MISSING** |
+| `showTokens` state       | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+| `contextLimit` memo      | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+| IN/OUT token display     | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+| Reasoning token display  | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+| "Toggle tokens" command  | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+| User message token count | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx` | **MISSING** |
+
+### Backend Token Utility (Exists)
+
+```typescript
+// packages/opencode/src/util/token.ts
+Token.estimate(input: string)                    // Character-based estimation
+Token.toCharCount(tokenEstimate: number)         // Convert tokens to chars
+Token.toTokenEstimate(charCount: number)         // Convert chars to tokens
+Token.calculateToolResultTokens(parts)           // Estimate tool result size
+```
+
+### Message Schema Fields (Exist)
+
+```typescript
+// UserMessage
+sentEstimate: z.number().optional()
+contextEstimate: z.number().optional()
+
+// AssistantMessage
+outputEstimate: z.number().optional()
+reasoningEstimate: z.number().optional()
+contextEstimate: z.number().optional()
+sentEstimate: z.number().optional()
+```
+
+## Technical Approach
+
+### Token Estimation Logic
+
+- Simple estimation based on character count using `CHARS_PER_TOKEN` constant
+- `calculateToolResultTokens` estimates size of tool inputs, outputs, and errors
+- Estimates prefixed with `~` to indicate they are approximate
+
+### Display Format
+
+- `IN X↓` - Input/context tokens (sent to model)
+- `OUT Y↑` - Output tokens (generated by model)
+- `~X think` - Reasoning tokens for thinking models
+- Context percentage: `X% of limit`
+
+## Implementation Tasks
+
+### Phase 1: Add Streaming Token Updates to Processor
+
+- [x] Import `Token` module in `packages/opencode/src/session/processor.ts`
+- [x] Add `reasoningTotal` and `textTotal` character accumulators at processor creation
+- [x] Update `reasoning-delta` handler to calculate and store `reasoningEstimate`:
+  ```typescript
+  case "reasoning-delta":
+    reasoningTotal += value.text.length
+    input.assistantMessage.reasoningEstimate = Token.toTokenEstimate(reasoningTotal)
+    await Session.updateMessage(input.assistantMessage)
+  ```
+- [x] Update `text-delta` handler to calculate and store `outputEstimate`:
+  ```typescript
+  case "text-delta":
+    textTotal += value.text.length
+    input.assistantMessage.outputEstimate = Token.toTokenEstimate(textTotal)
+    await Session.updateMessage(input.assistantMessage)
+  ```
+- [ ] Update `finish-step` to emit final `contextEstimate` from usage data
+
+### Phase 2: Add Token Display State
+
+- [x] Add `showTokens` signal to session component:
+  ```typescript
+  const [showTokens, setShowTokens] = createSignal(kv.get("show_tokens", false))
+  ```
+- [ ] Add `contextLimit` memo that gets limit from current model/provider
+- [x] Add to context provider: `showTokens: () => boolean`
+
+### Phase 3: Add Toggle Tokens Command
+
+- [x] Add "Toggle tokens" command to `command.register()` array:
+  ```typescript
+  {
+    title: showTokens() ? "Hide tokens" : "Show tokens",
+    value: "session.toggle.tokens",
+    category: "Session",
+    onSelect: (dialog) => {
+      setShowTokens((prev) => {
+        const next = !prev
+        kv.set("show_tokens", next)
+        return next
+      })
+      dialog.clear()
+    },
+  }
+  ```
+
+### Phase 4: Update AssistantMessage Component
+
+- [x] Add token calculation logic:
+  ```typescript
+  const inputTokens = createMemo(() => {
+    const sent = props.message.sentEstimate ?? 0
+    const context = props.message.contextEstimate ?? 0
+    return sent + context
+  })
+  const outputTokens = createMemo(() => props.message.tokens?.output ?? props.message.outputEstimate ?? 0)
+  const reasoningTokens = createMemo(() => props.message.tokens?.reasoning ?? props.message.reasoningEstimate ?? 0)
+  ```
+- [x] Add conditional token display:
+  ```tsx
+  <Show when={showTokens()}>
+    <text fg={theme.textMuted}>
+      IN {inputTokens().toLocaleString()}↓ OUT {outputTokens().toLocaleString()}↑
+      <Show when={reasoningTokens()}> ~{reasoningTokens().toLocaleString()} think</Show>
+    </text>
+  </Show>
+  ```
+
+### Phase 5: Update UserMessage Component
+
+- [x] Add individual token count display when `showTokens()` is true:
+  ```tsx
+  <Show when={showTokens() && props.message.sentEstimate}>
+    <text fg={theme.textMuted}>~{props.message.sentEstimate?.toLocaleString()} tokens</text>
+  </Show>
+  ```
+
+## Code References
+
+### Internal Files
+
+- `packages/opencode/src/util/token.ts` - Token utility functions (exists)
+- `packages/opencode/src/session/message-v2.ts:308-309` - User message estimate fields (exists)
+- `packages/opencode/src/session/message-v2.ts:369-372` - Assistant message estimate fields (exists)
+- `packages/opencode/src/session/processor.ts:82-88` - reasoning-delta handler (needs update)
+- `packages/opencode/src/session/processor.ts:305-315` - text-delta handler (needs update)
+- `packages/opencode/src/session/processor.ts:251-270` - finish-step handler (needs update)
+- `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx:1097-1162` - AssistantMessage component
+- `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx:1001-1095` - UserMessage component
+- `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx:241-777` - Command registration
+
+### External References
+
+- Original PR: https://github.com/sst/opencode/pull/4709
+
+## Estimated Changes
+
+| File           | Lines Added | Lines Modified |
+| -------------- | ----------- | -------------- |
+| `processor.ts` | ~15         | ~10            |
+| `index.tsx`    | ~50         | ~15            |
+| **Total**      | ~65         | ~25            |
+
+## Validation Criteria
+
+- [x] Token estimates display during streaming (before final usage available)
+- [x] `IN X↓` shows input/context tokens accurately
+- [x] `OUT Y↑` shows output tokens, updating in real-time during generation
+- [x] Reasoning tokens display for models that support thinking (e.g., Claude)
+- [x] "Toggle tokens" command appears in command palette
+- [x] Toggle persists via KV store across sessions
+- [x] User messages show estimated token count when toggle enabled
+- [x] Estimates use `~` prefix to indicate approximation
+- [x] Final token counts from API replace estimates when available
+
+## Dependencies
+
+None - all required utilities and schema fields already exist.
+
+## Risks & Considerations
+
+1. **Estimation Accuracy**: Character-based estimation is approximate. Actual tokenization varies by model. Consider this acceptable for UX purposes.
+
+2. **Performance**: Updating message on every delta may cause performance issues. Consider throttling updates (e.g., every 100ms or 100 chars).
+
+3. **Context Limit**: Different models have different context limits. Need to properly fetch limit from provider/model configuration.
+
+4. **Subtask Bug Status**: The regression bug mentioned in PR discussions (missing `updatePart` call after `taskTool.execute`) was previously fixed and the fix is still present. No action needed.

CONTEXT/PLAN-4791-restore-bash-viewer-ansi-2025-12-09.md

@@ -0,0 +1,188 @@
+# Plan: Restore Bash Tool Expansion & ANSI Output Feature (PR #4791)
+
+**Date:** 2025-12-09  
+**Related PR:** https://github.com/sst/opencode/pull/4791  
+**Status:** IMPLEMENTED - Feature restored 2025-12-10
+
+## Overview
+
+This plan documents the restoration of the "Bash Tool Expansion & Colored ANSI Output" feature that was originally added in PR #4791. The feature provides:
+
+- Full-screen viewer for bash command outputs
+- ANSI color rendering for terminal output
+- Output truncation in chat with "Click to view full output" button
+- Forced color output from CLI tools
+
+## Current State Analysis
+
+### What's Missing
+
+| Component                      | File                                                           | Status      |
+| ------------------------------ | -------------------------------------------------------------- | ----------- |
+| `ghostty-opentui` dependency   | `packages/opencode/package.json`                               | **MISSING** |
+| `ptyToText` import             | `packages/opencode/src/tool/bash.ts`                           | **MISSING** |
+| `FORCE_COLOR` env vars         | `packages/opencode/src/tool/bash.ts`                           | **MISSING** |
+| `bashOutput` signal            | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| `showBashOutput` context       | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| Full-screen bash viewer        | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| `ghostty-terminal` component   | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| Keyboard navigation for viewer | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| Bash tool truncated preview    | `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx`   | **MISSING** |
+| `initialValue` prop            | `packages/opencode/src/cli/cmd/tui/component/prompt/index.tsx` | **MISSING** |
+| `text` getter on PromptRef     | `packages/opencode/src/cli/cmd/tui/component/prompt/index.tsx` | **MISSING** |
+
+### Current Bash Tool Rendering
+
+The current bash tool simply strips ANSI codes and displays plain text:
+
+```typescript
+const output = createMemo(() => stripAnsi(props.metadata.output?.trim() ?? ""))
+// ...
+<text fg={theme.text}>{output()}</text>
+```
+
+## Technical Approach
+
+### ANSI Color Rendering
+
+- Use `ghostty-opentui` package for terminal rendering
+- `GhosttyTerminalRenderable` component renders ANSI codes properly
+- `ptyToText()` processes raw PTY output
+
+### Environment Variables for Color Output
+
+Force CLI tools to produce colored output even when not in TTY:
+
+```typescript
+env: {
+  FORCE_COLOR: "3",
+  CLICOLOR: "1",
+  CLICOLOR_FORCE: "1",
+  TERM: "xterm-256color",
+  TERM_PROGRAM: "bash-tool",
+  PY_COLORS: "1",
+  ANSICON: "1",
+  // ... more
+}
+```
+
+### Full-Screen Viewer
+
+- Toggle between chat view and full-screen bash viewer
+- Keyboard navigation: ESC to close, Page Up/Down, Home/End for scrolling
+- Preserve prompt text when switching views
+
+## Implementation Tasks
+
+### Phase 1: Add Dependencies
+
+- [x] Add `ghostty-opentui` to `packages/opencode/package.json`
+  ```json
+  "ghostty-opentui": "1.3.6"
+  ```
+- [x] Run `bun install` to update lockfile
+
+### Phase 2: Update Bash Tool
+
+- [x] Add import to `packages/opencode/src/tool/bash.ts`:
+  ```typescript
+  import { ptyToText } from "ghostty-opentui"
+  ```
+- [x] Update spawn environment variables (around line 225):
+  ```typescript
+  env: {
+    ...process.env,
+    FORCE_COLOR: "3",
+    CLICOLOR: "1",
+    CLICOLOR_FORCE: "1",
+    TERM: "xterm-256color",
+    TERM_PROGRAM: "bash-tool",
+    PY_COLORS: "1",
+    ANSICON: "1",
+    NO_COLOR: undefined,
+  }
+  ```
+- [x] Wrap output with `ptyToText()` before returning
+
+### Phase 3: Update Session Index
+
+- [x] Add `BashOutputView` type:
+  ```typescript
+  type BashOutputView = {
+    command: string
+    output: () => string
+  }
+  ```
+- [x] Add `bashOutput` signal: `createSignal<BashOutputView | undefined>(undefined)`
+- [x] Add `showBashOutput` function to context
+- [x] Register `ghostty-terminal` component with opentui
+- [x] Add keyboard handlers for viewer navigation (ESC, PageUp/Down, Home/End)
+- [x] Add conditional rendering that switches between scrollbox and bash viewer
+- [x] Add `promptDraft` signal for preserving prompt text
+
+### Phase 4: Update Bash Tool Renderer
+
+- [x] Update the bash tool registration (around line 1382-1404):
+  - [x] Use `<ghostty-terminal>` for output preview
+  - [x] Limit preview to 20 lines
+  - [x] Add "Click to see full output" button when output exceeds limit
+  - [x] Wire click handler to `showBashOutput`
+
+### Phase 5: Update Prompt Component
+
+- [ ] Add `initialValue` prop to `PromptProps` type (line 29-36) - Not needed for basic implementation
+- [ ] Add `text` getter to `PromptRef` type (line 38-45) - Not needed for basic implementation
+- [ ] Handle `initialValue` in `onMount` to restore prompt text - Not needed for basic implementation
+
+## Code References
+
+### Internal Files
+
+- `packages/opencode/package.json` - Add ghostty-opentui dependency
+- `packages/opencode/src/tool/bash.ts:225-233` - spawn() call, needs env vars
+- `packages/opencode/src/tool/bash.ts:350-358` - return statement, needs ptyToText
+- `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx:80-89` - Context definition
+- `packages/opencode/src/cli/cmd/tui/routes/session/index.tsx:1382-1404` - Bash tool renderer
+- `packages/opencode/src/cli/cmd/tui/component/prompt/index.tsx:29-45` - PromptProps/PromptRef types
+
+### External References
+
+- Original PR: https://github.com/sst/opencode/pull/4791
+- ghostty-opentui package: https://www.npmjs.com/package/ghostty-opentui
+
+## Estimated Changes
+
+| File               | Lines Added | Lines Modified |
+| ------------------ | ----------- | -------------- |
+| `package.json`     | 1           | 0              |
+| `bash.ts`          | 15          | 5              |
+| `index.tsx`        | ~150        | ~30            |
+| `prompt/index.tsx` | 10          | 5              |
+| **Total**          | ~176        | ~40            |
+
+## Validation Criteria
+
+- [x] `bun install` succeeds with new dependency
+- [x] CLI tools produce colored output (test with `ls --color`, `git status`)
+- [x] Bash output in chat shows ANSI colors (not raw escape codes)
+- [x] Long outputs are truncated to 20 lines in chat preview
+- [x] "Click to see full output" button appears for truncated outputs
+- [x] Clicking opens full-screen bash viewer
+- [x] Full-screen viewer shows complete output with colors
+- [x] ESC key closes full-screen viewer
+- [x] Page Up/Down, Home/End work in viewer
+- [ ] Prompt text is preserved when opening/closing viewer - Minor, can be addressed later
+
+## Dependencies
+
+- `ghostty-opentui` npm package (needs to be added)
+
+## Risks & Considerations
+
+1. **Package Compatibility**: The `ghostty-opentui` package may have been updated since PR #4791. Check for any API changes.
+
+2. **Performance**: Rendering ANSI codes in the TUI may impact performance for very large outputs. The 20-line preview helps mitigate this.
+
+3. **Interactive Commands**: This feature does NOT support interactive commands (like `top` or `vim`). It's strictly for static output rendering.
+
+4. **Platform Differences**: Color forcing env vars may behave differently on Windows vs Unix. Test on multiple platforms.