overcode 0.5.8__tar.gz → 0.6.2__tar.gz

{overcode-0.5.8/src/overcode.egg-info → overcode-0.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: overcode
-Version: 0.5.8
+Version: 0.6.2
 Summary: A supervisor for managing multiple Claude Code instances in tmux
 Author: Mike Bond
 Project-URL: Homepage, https://github.com/mkb23/overcode
@@ -23,7 +23,7 @@ Requires-Dist: textual>=0.40.0
 Requires-Dist: rich>=13.0.0
 Requires-Dist: typer>=0.9.0
 Requires-Dist: pyyaml>=6.0
-Requires-Dist: libtmux>=0.37.0
+Requires-Dist: libtmux<0.63,>=0.37.0
 Provides-Extra: dev
 Requires-Dist: pytest>=7.0.0; extra == "dev"
 Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
@@ -67,19 +67,17 @@ Running multiple coding agents is powerful, but managing them gets chaotic fast.
 Try it instantly with [uvx](https://docs.astral.sh/uv/):
 
 ```bash
-uvx overcode monitor
+uvx overcode
 ```
 
-This opens the standalone dashboard. Press `n` to create your first agent.
-
-For the full tmux-native experience (recommended):
+Or install it:
 
 ```bash
 pip install overcode
-overcode tmux
+overcode            # same as `overcode tmux`
 ```
 
-This creates a split layout: the overcode dashboard on top, the focused agent's live terminal on the bottom. Navigate agents with `j/k` — the bottom pane follows automatically. Press `Tab` to toggle focus between panes.
+This opens the tmux split layout (attaching to it if you are in a plain terminal): the overcode dashboard on top, the focused agent's live terminal on the bottom. Press `n` to create your first agent. Navigate agents with `j/k` — the bottom pane follows automatically. Press `Tab` to toggle focus between panes.
 
 **Requirements:** Python 3.12+, tmux, and an agent CLI — [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [opencode](https://opencode.ai) (v1, or the opencode2 2.0 preview), [Codex CLI](https://github.com/openai/codex), [Grok Build](https://x.ai), or [Hermes Agent](https://github.com/NousResearch/hermes-agent)
 
@@ -88,7 +86,7 @@ See the [Getting Started Guide](docs/getting-started.md) for a complete walkthro
 ## Features
 
 ### Tmux Split Layout (`overcode tmux`)
-The recommended way to use overcode. Creates a two-pane layout in tmux:
+The overcode dashboard. `overcode`, `overcode tmux` and `overcode monitor` all open it. A two-pane layout in tmux:
 - **Top pane**: Compact agent dashboard with live status
 - **Bottom pane**: The focused agent's native terminal — no emulation, real tmux
 - `j/k` navigates agents, bottom pane follows
@@ -108,7 +106,7 @@ The TUI displays all agents with live status updates, showing:
 - **Launch agents** with custom prompts and permission settings
 - **Send instructions** directly from the dashboard
 - **Standing orders** - persistent instructions that guide agent behavior
-- **Agent hierarchy** - parent/child delegation with follow mode and reporting
+- **Agent hierarchy** - parent/child delegation with follow mode, reporting and oversight presets (see [Child Agents](docs/children.md))
 - **Cost budgets** - per-agent spending limits with automatic enforcement
 - **Sleep mode** - pause agents and exclude them from stats
 - **Shutdown and revive**: `overcode shutdown` stops agents, jobs, daemons and tmux sessions and keeps the agent records. After a reboot, `overcode revive --all` brings the agents back with their conversations.
@@ -146,11 +144,10 @@ limitations.
 Aggregate agents from multiple machines into one dashboard:
 - Configure sister machines in `~/.overcode/config.yaml`
 - Remote agents appear alongside local ones
-- In tmux split mode, selecting a sister agent auto-zooms the dashboard with a preview pane
+- Selecting a sister agent auto-zooms the dashboard with a preview pane
+- Each remote runs `overcode web`, a small API server that serves the sister API only (there is no web UI)
 
 ### Analytics & Export
-- **Web dashboard** - mobile-friendly monitoring from any device
-- **Historical analytics** - browse session history with charts
 - **Parquet export** - analyze data in Jupyter notebooks
 - **Presence tracking** - correlate activity with your availability
 

{overcode-0.5.8 → overcode-0.6.2}/README.md

@@ -25,19 +25,17 @@ Running multiple coding agents is powerful, but managing them gets chaotic fast.
 Try it instantly with [uvx](https://docs.astral.sh/uv/):
 
 ```bash
-uvx overcode monitor
+uvx overcode
 ```
 
-This opens the standalone dashboard. Press `n` to create your first agent.
-
-For the full tmux-native experience (recommended):
+Or install it:
 
 ```bash
 pip install overcode
-overcode tmux
+overcode            # same as `overcode tmux`
 ```
 
-This creates a split layout: the overcode dashboard on top, the focused agent's live terminal on the bottom. Navigate agents with `j/k` — the bottom pane follows automatically. Press `Tab` to toggle focus between panes.
+This opens the tmux split layout (attaching to it if you are in a plain terminal): the overcode dashboard on top, the focused agent's live terminal on the bottom. Press `n` to create your first agent. Navigate agents with `j/k` — the bottom pane follows automatically. Press `Tab` to toggle focus between panes.
 
 **Requirements:** Python 3.12+, tmux, and an agent CLI — [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [opencode](https://opencode.ai) (v1, or the opencode2 2.0 preview), [Codex CLI](https://github.com/openai/codex), [Grok Build](https://x.ai), or [Hermes Agent](https://github.com/NousResearch/hermes-agent)
 
@@ -46,7 +44,7 @@ See the [Getting Started Guide](docs/getting-started.md) for a complete walkthro
 ## Features
 
 ### Tmux Split Layout (`overcode tmux`)
-The recommended way to use overcode. Creates a two-pane layout in tmux:
+The overcode dashboard. `overcode`, `overcode tmux` and `overcode monitor` all open it. A two-pane layout in tmux:
 - **Top pane**: Compact agent dashboard with live status
 - **Bottom pane**: The focused agent's native terminal — no emulation, real tmux
 - `j/k` navigates agents, bottom pane follows
@@ -66,7 +64,7 @@ The TUI displays all agents with live status updates, showing:
 - **Launch agents** with custom prompts and permission settings
 - **Send instructions** directly from the dashboard
 - **Standing orders** - persistent instructions that guide agent behavior
-- **Agent hierarchy** - parent/child delegation with follow mode and reporting
+- **Agent hierarchy** - parent/child delegation with follow mode, reporting and oversight presets (see [Child Agents](docs/children.md))
 - **Cost budgets** - per-agent spending limits with automatic enforcement
 - **Sleep mode** - pause agents and exclude them from stats
 - **Shutdown and revive**: `overcode shutdown` stops agents, jobs, daemons and tmux sessions and keeps the agent records. After a reboot, `overcode revive --all` brings the agents back with their conversations.
@@ -104,11 +102,10 @@ limitations.
 Aggregate agents from multiple machines into one dashboard:
 - Configure sister machines in `~/.overcode/config.yaml`
 - Remote agents appear alongside local ones
-- In tmux split mode, selecting a sister agent auto-zooms the dashboard with a preview pane
+- Selecting a sister agent auto-zooms the dashboard with a preview pane
+- Each remote runs `overcode web`, a small API server that serves the sister API only (there is no web UI)
 
 ### Analytics & Export
-- **Web dashboard** - mobile-friendly monitoring from any device
-- **Historical analytics** - browse session history with charts
 - **Parquet export** - analyze data in Jupyter notebooks
 - **Presence tracking** - correlate activity with your availability
 

{overcode-0.5.8 → overcode-0.6.2}/docs/README.md

@@ -11,6 +11,7 @@ Welcome to the overcode documentation. Overcode is a TUI supervisor for managing
 | [CLI Reference](cli-reference.md) | Complete reference for all commands and options |
 | [Configuration](configuration.md) | Config file, environment variables, and presets |
 | [Advanced Features](advanced-features.md) | Sleep mode, remote monitoring, and more |
+| [Child Agents](children.md) | Parent/child trees, reports, oversight presets, budgets, and when to use a child instead of a subagent |
 | [Skill Profiles](skill-profiles.md) | Named sets of skills switched on per agent, instead of always on |
 | [Agent Backends](backends.md) | Running opencode/Codex/Grok agents; support matrix and flag mapping |
 | [Claude Session Files](claude-session-files.md) | How token counting and cost estimation works internally |

{overcode-0.5.8 → overcode-0.6.2}/docs/advanced-features.md

@@ -136,11 +136,30 @@ Heartbeats and supervision resume automatically on the next daemon loop (~10s).
 
 ### No Global Default
 
-Budgets are per-agent only. There is no global default budget in config.
+Budgets are per-agent only. There is no global default budget in config;
+a [children preset](configuration.md#children-presets) can give the
+children launched with it a default budget.
+
+## Billing Codes
+
+Every agent's spend is filed under a billing code in a ledger that is kept
+for ever: its launch name by default (all the runs named `shirka` add up
+under `shirka`), or `overcode launch --billing-code client-x`. A child or a
+fork always takes its parent's code, so a whole tree of delegated work adds
+up under the code its root was launched with. `overcode rebill <agent>
+<code>` moves an agent to another code from now on (its spend so far stays
+where it was), and `overcode billing` prints spend per code (`--since 30d`,
+`--month 2026-09`). A code has a fixed id and a name you can change:
+`overcode billing rename` renames it in every report, past ones included,
+and `overcode billing merge` folds codes into another (several at once, or a quoted pattern). In the TUI, the new-agent dialog and the "Billing
+code…" palette command cycle a `billing.codes` shortlist from config.yaml
+with `Space` ([TUI guide](tui-guide.md#billing-codes)). Codes have no
+limits of their own: budgets stay per agent. Details:
+[Billing Codes](configuration.md#billing-codes).
 
 ## Agent Hierarchy
 
-Agents can spawn child agents, creating a parent/child tree. This enables delegation patterns where a parent agent launches children to handle subtasks.
+Agents can spawn child agents, creating a parent/child tree. This enables delegation patterns where a parent agent launches children to handle subtasks. [Child Agents](children.md) covers the whole model, including presets and when to use a child instead of a Claude subagent.
 
 ### Launching Child Agents
 
@@ -169,7 +188,7 @@ overcode launch -n child --follow -p "Do the task. When done: overcode report --
 overcode follow child-agent
 ```
 
-When the child reports success, it's marked "done" — it stays alive in tmux but is hidden from default views.
+When the child reports success, it's marked "done" — it stays alive in tmux but is hidden from default views, until the monitor daemon closes its window after its archive delay (1 h by default; `--archive-after` or a preset changes it).
 
 ### Reporting Completion
 
@@ -180,7 +199,7 @@ overcode report --status success --reason "All tests pass"
 overcode report --status failure --reason "Build failed"
 ```
 
-Without a report, the child enters `waiting_oversight` status (👁️) when it stops. The parent's follow mode blocks until a report arrives or the oversight policy triggers.
+Without a report, the child enters `waiting_oversight` status when it stops: at rest (⚪), with 📭 as its result (the TUI's `l` result mode). The parent's follow mode blocks until a report arrives or the oversight policy triggers.
 
 **Always include `overcode report` at the end of your `--prompt`.** This is the most common mistake when setting up delegation.
 
@@ -199,7 +218,9 @@ overcode launch -n child --follow --on-stuck fail -p "..."
 overcode launch -n child --follow --oversight-timeout 5m -p "..."
 ```
 
-The TUI shows a countdown timer (⏳) for agents with a timeout policy.
+`--preset supervised|balanced|autonomous` sets the oversight policy and the archive delay together (see [Presets](children.md#presets)).
+
+The TUI shows a countdown timer (⏳) for agents with a timeout policy. The monitor daemon starts the timeout when the child stops, so it applies whether or not a parent follows it; a child prompted again before it expires loses its deadline and gets a fresh one when it next stops. An expired timeout shows ⌛ as the result.
 
 ### Budget Transfer
 
@@ -230,28 +251,29 @@ Press `S` to cycle to "Tree" sort mode. Agents display in a hierarchy with tree
 parent-agent    🟢  2.5h  ...
 ├─child-1       🟢  1.2h  ...
 │ └─grandchild  🟠  0.5h  ...
-└─child-2       ✓   0.8h  ...
+└─child-2       ⚪   0.8h  ✅ all tests pass ...
 ```
 
-- **`X`** — Collapse/expand children of the focused agent
-- **`D`** — Show/hide "done" child agents (hidden by default)
+- **`←` / `→`** — Fold / unfold the focused agent: its children, 🌙 background sessions and delegated work together. `→` on an open agent steps onto its first child; `←` on a closed one moves to its parent ([TUI guide → Folding](tui-guide.md#folding))
+- **`X`** — Close everything open, or with nothing open, open everything
+- **`d`** (Agent visibility) — Show/hide "done" child agents (hidden by default)
 
-The child count column (👶) shows how many direct children each agent has.
+The child count column (👶) shows how many direct children each agent has, then how they stand: `👶 5 2▶ 1📭 1✅ 1❌` (▶ still working, 📭 stopped without a report, ✅ success, ❌ failure or timeout; zero counts are left out). A parent whose turn ended while its children still work shows 🟡 ("Waiting on N children"), not 🔴.
 
 ### Done Agents
 
-When a child reports success or failure via `overcode report`, it enters the "done" state:
+When a child reports success or failure via `overcode report`, it enters the "done" state, at rest (⚪) with ✅ or ❌ and the reason as its result (the TUI's `l` result mode; `overcode list` prints it in full):
 - Hidden from default list and TUI views
-- Press `D` in TUI or use `--show-done` in CLI to reveal them
+- Turn on "Done child agents" in the TUI's Agent visibility dialog (`d`), or use `--show-done` in CLI, to reveal them
 - Auto-archived by the monitor daemon after 1 hour
 - Manually archive with `overcode cleanup --done`
 
 ### Waiting for Oversight
 
-If a child stops without calling `overcode report`, it enters `waiting_oversight` (👁️):
-- Visible in the TUI with a yellow indicator
+If a child stops without calling `overcode report`, it enters `waiting_oversight`:
+- Visible in the TUI at rest (⚪), with 📭 as its result; it never rings the bell
 - Parent's follow mode remains blocked, waiting for a report
-- With `--oversight-timeout`, a countdown timer appears in the TUI
+- With `--oversight-timeout`, a countdown timer appears in the TUI (the daemon enforces it, with or without `--follow`)
 - Resolve by attaching to the child's tmux window and running `overcode report` manually, or let the timeout expire
 
 ### Filtered List
@@ -298,37 +320,42 @@ Quickly reset an agent to the main branch.
 - Starting fresh on a new task
 - Pulling in changes from other agents
 
-## Remote Monitoring
-
-Monitor agents from anywhere with the web dashboard.
-
-### Web Dashboard
+## API Server
 
-Start the dashboard server:
+`overcode web` runs a small HTTP API server. It serves the **sister API only** —
+what another machine's overcode calls when it adds this one as a sister (see
+[Sister Integration](#sister-integration)). There is no web UI: the HTML
+dashboard, analytics pages and the Cloudflare relay were removed.
 
 ```bash
 overcode web                          # Start on localhost:8080
-overcode web --host 0.0.0.0           # LAN access (needs api_key)
+overcode web --host 0.0.0.0           # LAN access (refused without web.api_key)
 overcode web --port 3000              # Custom port
 overcode web --stop                   # Stop the server
 ```
 
 The server runs in the background — the command exits immediately. Run `overcode web` again to see the URL if already running.
 
-Access `http://<your-ip>:8080` from your phone, tablet, or another machine.
+Routes (all JSON; `X-API-Key` required when `web.api_key` is set):
 
-Features:
-- Summary statistics and daily activity charts
-- Session browser with sorting
-- Timeline visualization with agent status and user presence
-- Efficiency metrics and cost analysis
-- Live agent monitoring at `/dashboard`
-- Sister-compatible `/api/status` endpoint
-- Auto-refreshes, optimized for mobile screens
+| Route | Used by |
+|-------|---------|
+| `GET /api/status` | sister TUI poll, SSH provisioner version check |
+| `GET /api/agents/{name}/status` | sister TUI focused-agent poll |
+| `GET /api/timeline/raw` | sister TUI timeline |
+| `GET /health` | SSH provisioner |
+| `POST/PUT/DELETE /api/agents/...`, `/api/daemon/...` | sister control actions (needs `web.allow_control: true` and `web.api_key`) |
+| `POST /api/shutdown` | SSH provisioner (stops services, agents keep running) |
+
+The server sends no CORS headers: its clients are overcode processes, not
+browsers, so a web page you visit cannot read your fleet through it. Binding
+off localhost needs `web.api_key`, and `web.allow_control` does too (control
+routes answer 403 without a key). Sisters send the key as `X-API-Key`; set
+the same `api_key` on the sister's entry in `sisters:`.
 
 ### TUI Toggle
 
-Press `w` in the TUI to start/stop the web server. The URL appears in the daemon panel.
+Press `w` in the TUI to start/stop the API server. Its port appears in the daemon status bar.
 
 ## AI Summarizer
 
@@ -463,38 +490,6 @@ Presence tracking works on all platforms. Sleep detection (gap inference), TUI h
 ~/.overcode/presence_log.csv
 ```
 
-### Analytics
-
-The web dashboard (`overcode web`) overlays presence data on timelines, showing when agents were working while you were away vs. actively monitoring.
-
-## Cloud Relay
-
-Push status to a remote endpoint for custom integrations.
-
-### Configuration
-
-```yaml
-relay:
-  enabled: true
-  url: https://your-worker.workers.dev/update
-  api_key: your-secret-key
-  interval: 30  # seconds
-```
-
-### Payload
-
-The relay sends JSON with:
-- All agent statuses
-- Token usage
-- Costs
-- Current activity
-
-### Use Cases
-
-- Custom dashboards
-- Slack notifications
-- Integration with project management tools
-
 ## Sister Integration
 
 Aggregate agents from multiple machines into a single TUI. When running overcode on 2-3 machines simultaneously, sister integration gives you read-only visibility into all agents from one place.
@@ -510,7 +505,10 @@ overcode web
 
 ### Setup Option A: SSH Tunnels (Recommended)
 
-SSH tunnels keep all traffic encrypted with no API key needed.
+SSH tunnels keep all traffic encrypted with no API key needed for viewing.
+To also control the remote's agents from here, the remote needs
+`web.allow_control: true` and a `web.api_key` (control routes answer 403
+without one), and the sister entry below needs the same `api_key`.
 
 On the **monitoring machine**, add tunnels in `~/.ssh/config`:
 
@@ -545,7 +543,7 @@ sisters:
 
 ### Setup Option B: Direct LAN with API Key
 
-Expose the web server directly on the LAN. Requires an API key for security.
+Expose the API server directly on the LAN. Requires an API key for security.
 
 On each **remote machine**, add to `~/.overcode/config.yaml`:
 
@@ -616,7 +614,7 @@ This typically happens if:
 For debugging TUI issues:
 
 ```bash
-overcode monitor --diagnostics
+overcode monitor --diagnostics   # opens the split with a diagnostics-mode dashboard
 ```
 
 This disables auto-refresh timers, letting you manually refresh with `r` and observe state changes.
@@ -625,7 +623,7 @@ This disables auto-refresh timers, letting you manually refresh with `r` and obs
 
 ### Color Issues
 
-If colors render incorrectly, press `M` for monochrome mode. This strips ANSI codes from the preview pane.
+If colors render incorrectly, press `M` for monochrome mode. This strips ANSI codes from the preview pane (sister agents and jobs).
 
 ### Mouse Issues
 
@@ -634,7 +632,3 @@ If mouse events interfere with your terminal:
 2. Mouse capture is disabled
 3. Select text normally
 4. Press `y` again to re-enable
-
-### Large Repos
-
-For repos with many files, Claude's output can be verbose. Use `v` to cycle through detail line counts (5/10/20/50) to manage preview pane size.

{overcode-0.5.8 → overcode-0.6.2}/docs/architecture.md

@@ -18,7 +18,7 @@ header-includes: |
 
 # Overview
 
-Overcode is a CLI/TUI orchestration layer for managing multiple Claude Code agent sessions running in tmux. It provides real-time status monitoring, time tracking, cost estimation, heartbeat automation, and a web dashboard — all built on top of tmux as the process host and Claude Code's JSONL history as the data source.
+Overcode is a CLI/TUI orchestration layer for managing multiple Claude Code agent sessions running in tmux. It provides real-time status monitoring, time tracking, cost estimation, heartbeat automation, and an HTTP API for sister instances — all built on top of tmux as the process host and Claude Code's JSONL history as the data source.
 
 **The codebase is 28,950 lines of Python across 72 source files.**
 
@@ -66,7 +66,7 @@ The core data pipeline is simple: tmux panes are scraped for status, the monitor
 
 3. **Consumers** read the published state file:
    - **TUI** (Textual, ~2s refresh) — merges daemon state + session registry + sister data
-   - **Web Dashboard** (stdlib HTTP) — serves JSON API + HTML dashboard
+   - **API Server** (stdlib HTTP) — serves the JSON sister API (no HTML UI since the web UI was removed)
    - **Supervisor Daemon** — reads state to decide when to launch intervention agents
 
 \newpage
@@ -93,6 +93,8 @@ The largest subsystem, split across:
 
 The TUI already has good separation between pure logic and UI code. The action mixin pattern keeps `tui.py` from being even larger, though the mixins themselves are tightly coupled to `SupervisorTUI`'s internal state.
 
+Since 0.6.0 the TUI has one layout: it is the top pane of the `overcode tmux` split, and navigating agents switches the window of a linked tmux session shown in the bottom pane. `overcode`, `overcode monitor` and `overcode tmux` all build that split (`cli/split.py` `open_split`), which runs `overcode monitor --sync-target <linked session>` in its top pane. The standalone (non-split) dashboard, its preview toggle, fullscreen preview and manual pane sync were removed; the preview pane remains only for sister agents and jobs view. (Line counts on this page predate that change.)
+
 ## Status Detection — 1,481 lines
 
 Four modules implementing a dual-strategy pattern:
@@ -115,6 +117,11 @@ Four modules implementing a dual-strategy pattern:
 
 ## Web Server — 3,699 lines (13% of codebase)
 
+> **Update:** the HTML dashboard, analytics pages, `web_templates.py`,
+> `web_chartjs.py` and the Cloudflare relay have since been removed. The
+> server (`web_server.py`, `web_api.py`, `web_control_api.py`) now serves the
+> sister API only. The figures below are from the original analysis.
+
 - `web_templates.py` (1,656 lines): **the single biggest contributor to bloat** — entire HTML/CSS/JS dashboard as Python string literals
 - `web_server.py` (642 lines): stdlib `http.server` routing
 - `web_api.py` (844 lines): data aggregation for JSON API endpoints
@@ -190,15 +197,15 @@ predates this reports nothing and is read as claude-code with everything on.
   `GrokStatsReader` reads a full, genuinely billing-accurate token/cost split
   from `updates.jsonl`; `NullStatsReader` answers "unknown" for backends with
   neither.
-- *Telemetry* — the hook-state file format (`hook_state_<agent>.json`,
-  `hook_events_<agent>.jsonl`) is overcode's own neutral schema. Claude Code
+- *Telemetry* — the hook-state file format (`agents/<session id>/hook_state.json`,
+  `hook_events.jsonl`; name-keyed `hook_state_<agent>.json` before 0.6) is overcode's own neutral schema. Claude Code
   fills it via `--settings` hooks calling `overcode hook-handler`; opencode
   fills it from the bundled plugin; codex fills it from per-launch `-c
   'hooks.<Event>=[...]'` config overrides plus
   `--dangerously-bypass-hook-trust` (snake_case stdin, close enough to
   Claude's own that little dialect translation is needed); grok fills it from
   a global, inert-outside-overcode `~/.grok/hooks/overcode.json` file
-  (camelCase stdin, translated by `hook_handler._normalize_hook_payload()`).
+  (camelCase stdin, translated by `hook_dialects.GrokHookDialect`).
   Any backend that can write those files gets the full `HookStatusDetector`
   experience for free.
 
@@ -435,7 +442,6 @@ Replace the four separate files (daemon state JSON, sessions JSON, status histor
 | `tui_actions/input.py` | 217 | TUI |
 | `tui_actions/daemon.py` | 201 | TUI |
 | `help_overlay.py` | 197 | TUI |
-| `logging_config.py` | 193 | Infra |
 | `protocols.py` | 189 | Tmux |
 | `testing/tui_eye.py` | 185 | Testing |
 | `claude_config.py` | 186 | Claude |
@@ -443,11 +449,9 @@ Replace the four separate files (daemon state JSON, sessions JSON, status histor
 | `bundled_skills.py` | 175 | Claude |
 | `daemon_panel.py` | 169 | TUI |
 | `summary_config_modal.py` | 165 | TUI |
-| `mocks.py` | 156 | Tmux |
 | `notifier.py` | 145 | Infra |
 | `daemon_logging.py` | 144 | Infra |
 | `summary_groups.py` | 141 | TUI |
-| `fullscreen_preview.py` | 130 | TUI |
 | `tmux_utils.py` | 128 | Tmux |
 | `hook_handler.py` | 123 | Claude |
 | `usage_monitor.py` | 119 | Infra |
@@ -457,5 +461,4 @@ Replace the four separate files (daemon state JSON, sessions JSON, status histor
 | `daemon_utils.py` | 93 | Infra |
 | `preview_pane.py` | 101 | TUI |
 | `status_detector_factory.py` | 83 | Status |
-| `interfaces.py` | 49 | Tmux |
 | `web_chartjs.py` | 32 | Web |