refactor: remove dead code, deprecated v1 schema, and fix stale docs (#6)

* refactor: remove dead code, deprecated v1 schema, and fix stale docs

Remove 4 dead functions from legacy character-based parser (jsonl.py),
dead get_all_widgets(), 8 dead dataclass fields across TokenMetrics/
ContextWindow/RenderContext, deprecated StatusLineConfig v1 schema class,
and 2 dead constants. Simplify vestigial UUID guard in debug.py. Fix
extensive README/AGENTS.md documentation rot: remove nonexistent
subscription widget, calibration tool section, v1 config examples, and
stale field/worker claims. Consolidate script/ into scripts/.

* fix: address review feedback on dead code cleanup

Fix benchmark.sh self-references after script→scripts rename.
Strengthen test assertions for image token counting and compact
boundary behavior.

* fix: add PSR build_command to sync lockfile on release

Author: wpfleger96

AGENTS.md

@@ -90,9 +90,9 @@ tests/
 
 **Widget System**: All widgets extend `Widget` base class with `render(context: RenderContext) -> Optional[str]`. Register with `@register_widget(display_name, default_color, description, fallback_text)`.
 
-**RenderContext**: Dataclass passed to all widgets containing `data`, `token_metrics`, `git_status`, `session_metrics`, `subscription_info`, `context_window`.
+**RenderContext**: Dataclass passed to all widgets containing `data`, `token_metrics`, `git_status`, `duration_seconds`, `terminal_width`, `context_window`.
 
-**Parallel I/O**: `statusline.py` uses `ThreadPoolExecutor(max_workers=3)` for transcript parsing, subscription loading, and model prefetch.
+**Parallel I/O**: `statusline.py` uses `ThreadPoolExecutor(max_workers=2)` for transcript parsing and model prefetch.
 
 **Config Caching**: `config/loader.py` caches config files by mtime to avoid re-reading unchanged files.
 
@@ -141,7 +141,7 @@ Fixtures in `tests/conftest.py`: `mock_stdin`, `sample_input_payload`, `basic_se
 | Modify token counting | `parsers/tokens.py`, `tests/unit/test_token_counting.py` |
 | Context window handling | `types.py`, `statusline.py`, `utils/models.py`, `tests/unit/test_context_window.py` |
 | Status line rendering | `renderer.py`, `config/defaults.py` |
-| Config schema changes | `config/schema.py`, `config/loader.py`, `tests/integration/test_config.py` |
+| Config schema changes | `config/schema.py`, `config/loader.py`, `tests/integration/test_config_system.py` |
 | CLI commands (install/doctor) | `cli/commands.py`, `statusline.py` (argparse setup) |
 | Git status display | `utils/git.py`, `widgets/builtin/git.py` |
 | Installation automation | `scripts/install.sh`, `scripts/uninstall.sh` |

README.md

@@ -131,25 +131,20 @@ This checks:
 Customize your status line by creating `~/.config/claude-statusline/config.yaml`:
 
 ```yaml
-version: 1
-lines:
-  - - type: model
-      color: cyan
-    - type: separator
-    - type: directory
-      color: blue
-    - type: separator
-    - type: git-branch
-      color: magenta
-    - type: separator
-    - type: context-percentage
-      color: auto  # auto-colors based on usage
-    - type: separator
-    - type: cost
-      color: auto  # auto-colors based on amount
-    - type: separator
-    - type: session-clock
-      color: white
+version: 2
+widgets:
+  model:
+    color: cyan
+  directory:
+    color: blue
+  git-branch:
+    color: magenta
+  context-percentage:
+    color: auto  # auto-colors based on usage
+  cost:
+    color: auto  # auto-colors based on amount
+  session-clock:
+    color: white
 ```
 
 **Widget Options:**
@@ -160,15 +155,8 @@ lines:
 
 **Color Options:** `white`, `black`, `red`, `green`, `yellow`, `blue`, `magenta`, `cyan`, `dim`, `auto`, `none`
 
-**Auto-Generated Config:**
-If no config file exists, one is automatically created with all widgets enabled. Delete `~/.config/claude-statusline/config.yaml` to regenerate defaults.
-
-**Stale Config Warning:**
-When new widgets are added, the script warns you:
-```
-Warning: Config is missing widgets from defaults: session-clock.
-Delete ~/.config/claude-statusline/config.yaml to regenerate with new defaults.
-```
+**Default Config:**
+If no config file exists, all default widgets are shown with built-in defaults. Create `~/.config/claude-statusline/config.yaml` to customize colors, ordering, or disable specific widgets.
 
 ## Features
 
@@ -177,11 +165,9 @@ Delete ~/.config/claude-statusline/config.yaml to regenerate with new defaults.
 - **Built-in Widgets**: 14 widgets including model, directory, git status, context, cost, session info
 - **Flexible Styling**: Per-widget color customization and bold formatting
 - **Automatic Fallback**: Missing config uses sensible defaults with all widgets enabled
-- **Stale Config Detection**: Warns when config is missing newly-added widgets
 
 **Available Widgets:**
 - `model` - Claude model name (e.g., "Sonnet 4.5")
-- `subscription` - Subscription type (Pro/Max) or API usage indicator
 - `directory` - Current working directory
 - `git-branch` - Active git branch name
 - `git-changes` - Staged/unstaged changes count
@@ -207,7 +193,6 @@ Delete ~/.config/claude-statusline/config.yaml to regenerate with new defaults.
 - **Priority-Based Sources**: Prefers `context_window` payload data → transcript parsing → model lookups
 - **Session Metrics**: Continues parsing transcripts for session duration and compact boundary detection
 - **Backwards Compatible**: Gracefully handles missing or null `context_window` data from older Claude Code versions
-- **Calibration Tool**: Includes `calibrate_token_counting.py` to validate and improve accuracy
 
 ### Token Calculation
 
@@ -343,60 +328,6 @@ Debug logs are written to **per-session files** in the `logs/` directory:
 
 ---
 
-## Token Counting Calibration Tool
-
-The project includes `claude-calibrate`, a command-line tool to verify and improve token counting accuracy against Claude's official measurements.
-
-### Purpose
-- Compare script calculations against Claude's official `/context` command output
-- Identify discrepancies and suggest calibration factors
-- Validate token counting accuracy when new Claude Code versions are released
-
-### Usage
-
-#### Semi-Automated Mode (Default)
-```bash
-uv run --no-config claude-calibrate session1.jsonl session2.jsonl --verbose
-```
-The tool provides precise instructions for resuming each session and prompts you to enter the official token counts.
-
-#### Manual Override Mode
-```bash
-uv run --no-config claude-calibrate session1.jsonl session2.jsonl \
-  --known-tokens 17.5k 68k --verbose
-```
-Provide known token counts to skip automatic session resumption (useful for sessions that can't be resumed).
-
-#### Auto-Discovery Mode
-```bash
-uv run --no-config claude-calibrate --max-sessions 3 --verbose
-```
-Automatically finds recent session files from all Claude Code project directories and provides instructions for manual calibration.
-
-### Example Calibration Report
-```
-📊 Successfully calibrated 2 session(s)
-
-INDIVIDUAL RESULTS:
-❌ session1.jsonl: Script 46,520 vs Claude 68,000 tokens (+31.6% difference)
-✅ session2.jsonl: Script 17,470 vs Claude 17,500 tokens (+0.2% difference)
-
-SUMMARY:
-Average discrepancy: +15.9% | Suggested calibration factor: 1.231
-⚠️  Token counting has moderate accuracy - consider adjusting CHARS_PER_TOKEN ratio
-```
-
-### Features
-- **Auto-discovery**: Finds recent sessions from all Claude Code project directories
-- **Semi-automated**: Provides precise resumption instructions for manual token collection
-- **Directory decoding**: Handles hyphenated paths (`-Users-name-Project` → `/Users/name/Project`)
-- **Flexible input**: Accepts raw numbers, K-suffixed values, or full `/context` output
-- **Accuracy validation**: Compares against Claude's official measurements
-- **Calibration recommendations**: Suggests specific improvements and correction factors
-- **uv compatibility**: Uses `uv run` for portable execution across Python environments
-
----
-
 ## Development
 
 ### Setup
@@ -405,19 +336,19 @@ Average discrepancy: +15.9% | Suggested calibration factor: 1.231
    ```bash
    git clone https://github.com/wpfleger96/claude-code-status-line.git
    cd claude-code-status-line
-   uv sync --no-config
+   uv sync
    ```
 
 2. **Configure Claude Code to use local version:**
    ```bash
-   uv run --no-config claude-statusline install
+   uv run claude-statusline install
    ```
 
 ### Testing Changes
 
 ```bash
 # Test statusline output
-echo '{"workspace": {"current_dir": "/test"}, "transcript_path": "", "model": {"id": "test", "display_name": "Test"}, "cost": {}, "version": "test"}' | uv run --no-config claude-statusline
+echo '{"workspace": {"current_dir": "/test"}, "transcript_path": "", "model": {"id": "test", "display_name": "Test"}, "cost": {}, "version": "test"}' | uv run claude-statusline
 
 # Run test suite
 uv run pytest

pyproject.toml

@@ -54,6 +54,7 @@ branch = "main"
 commit_parser = "conventional"
 commit_message = "chore(release): {version}"
 allow_zero_version = true
+build_command = "uv sync"
 
 [tool.semantic_release.commit_parser_opts]
 allowed_types = ["feat", "fix", "chore", "ci", "docs", "style", "refactor", "test"]

script/benchmark.sh scripts/benchmark.shscript/benchmark.sh renamed to scripts/benchmark.sh

@@ -56,10 +56,10 @@ Commands:
   help     - Show this help message
 
 Typical workflow:
-  1. ./script/benchmark.sh save          # Initial baseline
+  1. ./scripts/benchmark.sh save          # Initial baseline
   2. [make changes]
-  3. ./script/benchmark.sh compare       # See if faster (iterate)
-  4. ./script/benchmark.sh record        # Save the good result
+  3. ./scripts/benchmark.sh compare       # See if faster (iterate)
+  4. ./scripts/benchmark.sh record        # Save the good result
 EOF
         ;;
 esac

src/claude_code_statusline/config/defaults.py

@@ -1,29 +1,17 @@
 """Default configuration for Claude Code Status Line."""
 
-from .schema import StatusLineConfig, WidgetConfigModel
+from .schema import WidgetConfigModel
 
 
-def get_default_config() -> StatusLineConfig:
-    """Generate the default status line configuration."""
-    return StatusLineConfig(
-        version=1,
-        lines=[
-            [
-                WidgetConfigModel(type="model", color="cyan"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="directory", color="blue"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="git-branch", color="magenta"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="context-percentage"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="cost"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="lines-changed"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="session-id"),
-                WidgetConfigModel(type="separator"),
-                WidgetConfigModel(type="session-clock"),
-            ]
-        ],
-    )
+def get_default_widgets() -> list[WidgetConfigModel]:
+    """Return the default ordered widget list (separators excluded)."""
+    return [
+        WidgetConfigModel(type="model", color="cyan"),
+        WidgetConfigModel(type="directory", color="blue"),
+        WidgetConfigModel(type="git-branch", color="magenta"),
+        WidgetConfigModel(type="context-percentage"),
+        WidgetConfigModel(type="cost"),
+        WidgetConfigModel(type="lines-changed"),
+        WidgetConfigModel(type="session-id"),
+        WidgetConfigModel(type="session-clock"),
+    ]

src/claude_code_statusline/config/loader.py

@@ -9,7 +9,7 @@
 
 from pydantic import ValidationError
 
-from .defaults import get_default_config
+from .defaults import get_default_widgets
 from .schema import StatusLineConfigV2, WidgetConfigModel, WidgetOverride
 
 _cached_config: Optional[StatusLineConfigV2] = None
@@ -92,20 +92,20 @@ def get_effective_widgets() -> list[WidgetConfigModel]:
         List of widgets with separators interleaved, ready for rendering
     """
     config = load_config_file()
-    defaults = get_default_config()
+    default_widgets = get_default_widgets()
 
     if config.order:
         widget_types = config.order
     else:
-        widget_types = [w.type for w in defaults.lines[0] if w.type != "separator"]
+        widget_types = [w.type for w in default_widgets]
 
     widgets = []
     for wtype in widget_types:
         override = config.widgets.get(wtype, WidgetOverride())
         if not override.enabled:
             continue
 
-        default = next((w for w in defaults.lines[0] if w.type == wtype), None)
+        default = next((w for w in default_widgets if w.type == wtype), None)
         if default:
             widget = default.model_copy()
             if override.color:

src/claude_code_statusline/config/schema.py

@@ -17,15 +17,6 @@ class WidgetConfigModel(BaseModel):
     metadata: dict[str, str] = Field(default_factory=dict)
 
 
-class StatusLineConfig(BaseModel):
-    """Complete status line configuration (v1 - deprecated)."""
-
-    version: int = 1
-    lines: list[list[WidgetConfigModel]] = Field(default_factory=list)
-
-    model_config = {"extra": "forbid"}
-
-
 class WidgetOverride(BaseModel):
     """Override settings for a single widget."""