scopio

Code metrics auditor with SQLite history, diffs and quality gates.

Features

• Audits code metrics (lines, complexity, warnings, languages) across multiple projects.
• Stores audit history in SQLite with diffs and quality gates.
• Ingests external linter reports (Ruff, ESLint, Clippy, SARIF) — see Ingestion Cookbook.
• Exports results as JSON, CSV and Markdown.
• Compares audits and detects regressions for CI (--fail-on-regression).

Install

pip install scopio

Requirements

• Python >= 3.11
• lizard >= 1.24.0 (required for CSV parsing mode)
• scc >= 3.3.0 (optional, used for language/line counting) — the boyter/scc code counter, not the unrelated PyPI scc package:

  # Download a binary from https://github.com/boyter/scc/releases, or:
  go install github.com/boyter/scc/v3@latest
  

• git (optional, used for metadata)

Quick Start

scopio --config scopio.toml run

All output artifacts (database, JSON, CSV, Markdown) are stored in the .scopio/ directory by default. Use --output-dir to override.

If scc or lizard versions differ from the expected range, a non-blocking tool_version_diverge warning is emitted in the logs.

Usage

Run audit

scopio --config scopio.toml run
scopio --config scopio.toml run --verbose
scopio --config scopio.toml run --incremental

Report history

scopio report --project my-project --limit 10

Diff between audits

scopio diff --project my-project

Diff report with file-level detail

scopio diff-report --project my-project --files --threshold-ccn 6

CI integration

scopio ci --project my-project --fail-on-regression

Clean old audits

scopio clean --keep 50

Archive old data

scopio archive --older-than 2026-01-01 --format csv

Initialize config

scopio init

Configuration

[discovery]
ignore_hidden = true
projects = ["my-project"]

[filters]
global_dirs = ["node_modules", "target", "dist", "build"]
minified_files = ["*.min.js", "*.min.css"]
ignored_langs = ["JSON", "Markdown", "TOML"]

[quality_gates]
max_ccn = 10.0            # average CCN per function
max_function_ccn = 15.0   # worst function complexity (disabled if omitted)
max_warnings = 10
max_ccn_trend_increase = 0.2
trend_sensitive_projects = []

[quality_gates.per_project]
# "my-project" = { max_ccn = 12.0, max_warnings = 15, max_function_ccn = 20.0 }

[quality_gates.per_language]
# Python = { max_ccn = 12.0, max_warnings = 20 }

[ci]
max_loc_trend_increase = 0.0  # fail if LOC grows above this ratio (0.0 = any growth)

Audit behavior

• Each audit run creates/updates a single row per (project, branch, commit_hash).
• Same commit, same branch: metrics are upserted (updated) and runs_count is incremented. This avoids data staleness when re-running on the same commit (e.g. after changing .scopio.toml filters).
• Different commit: a new row is created with runs_count = 1.
• The runs_count field is exposed in report, CSV, JSON and Markdown exports.
• Diff and CI comparisons compare the previous audit against the current one by default (use --base first to compare against the very first audit).
• Quality gates use the current audit's values; the trend gate compares against the previous historical audit.

What the numbers mean

• LOC: source lines of code counted by scc.
• NLOC: logical lines of code counted by lizard. LOC and NLOC are intentionally different metrics (different counting methodologies).
• CCN (avg): cyclomatic complexity averaged over all functions. Averages hide outliers.
• ccn_max / max_function_ccn: cyclomatic complexity of the worst single function — this is what the per-function quality gate checks.
• warnings: currently always 0 — the lizard --csv format does not emit warnings. Language-specific warning adapters (clippy, eslint, ruff) are planned.
• file_metrics: stored per-function in the database; the granular diff-report aggregates them per file (NLOC sum, CCN max).

GitHub Actions

name: Scopio
on:
  push:
    branches: [main]
  pull_request:
    branches: [main]

jobs:
  scopio:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Install uv
        uses: astral-sh/setup-uv@v10.0.1
        with:
          enable-cache: true
          cache-dependency-glob: "uv.lock"
      - name: Install Python
        run: uv python install 3.11
      - name: Install scopio
        run: pip install scopio
      - name: Run scopio
        run: scopio ci --project my-project --fail-on-regression

Development

This project uses uv as the Python toolchain.

See CONTRIBUTING.md for the commit convention and release workflow.

Architecture and scope decisions are recorded in docs/DECISIONS.md.

# Sync dependencies and create a virtual environment
uv sync

# Run tests
uv run pytest

# Run a lint check
uv run ruff check

# Run scopio locally
uv run scopio --help

Semantic Versioning

This project adheres to Semantic Versioning.

License

MIT