AGENT SKILL

befund-complexity-score

Computes a relative complexity/attention-priority index per stage using the fixed formula 2.94 x (KSLOC)^1.10, and lists unmeasured stages plainly rather than inventing numbers. Use when the user asks "which module needs attention first", "score complexity per stage", "what's our tech debt hotspot", or as part of befund-autopilot's CHECK phase.

License 许可
MIT
所属仓库
Anselmoo/werkstoff
最近更新
2026-11-02
Stars
★ 1
Forks
0

01 前言 Front Matter

name
befund-complexity-score
description
Computes a relative complexity/attention-priority index per stage using the fixed formula 2.94 x (KSLOC)^1.10, and lists unmeasured stages plainly rather than inventing numbers. Use when the user asks "which module needs attention first", "score complexity per stage", "what's our tech debt hotspot", or as part of befund-autopilot's CHECK phase.

03 SKILL.md 正文

befund-complexity-score

Compute a relative complexity/attention-priority index per stage or module.

Step 0: Settings gate

python3 -B "${CLAUDE_PLUGIN_ROOT}/scripts/befund_cli.py" check-enabled --repo <repo_root> --skill befund-complexity-score

This skill has no Verify phase -- do not gate anything here on skip_verification; that setting simply does not apply to a pure measurement.

Step 1: Determine the stage list

Prefer <output_dir>/stage_map.json if it exists (reuse befund-stage-map's stages). Otherwise fall back to the detected-languages list from detect-languages, treating each language's file set as one pseudo-stage.

Step 2: Measure each stage

Dispatch complexity-surveyor per stage to measure SLOC, file count, and cyclomatic complexity (via whatever tool is available for that language -- radon, lizard, gocyclo, etc.). If no tool is available for a stage's language, the surveyor reports -1/0 per its own refusal list rather than fabricating a plausible-looking number -- when that happens, mark the stage unmeasured: true in the output and list it plainly in COMPLEXITY_SCORE.md instead of computing an index for it.

Step 3: Apply the exact formula -- no substitutes

Rule complexity-score-formula: the index MUST be 2.94 x (KSLOC)^1.10 and nothing else.

python3 -B "${CLAUDE_PLUGIN_ROOT}/scripts/befund_cli.py" complexity-index --ksloc <stage_sloc / 1000>

Use this for every measured stage rather than restating the constants inline -- the formula lives in one place (scripts/lib/formulas.py) so no per-skill copy can drift from it.

Step 4: Validate and write

python3 -B "${CLAUDE_PLUGIN_ROOT}/scripts/befund_cli.py" validate-artifact --kind complexity_score_summary --file <path-or-inline-json>

The validator recomputes 2.94 x (KSLOC)^1.10 for every non-unmeasured stage and rejects the artifact if the persisted complexity_index does not match to within floating-point tolerance -- a hand-typed or LLM-guessed number cannot pass.

A complexity_score_summary.json stage entry looks like this -- one measured, one not:

{
  "stages": [
    {"stage": "api", "sloc": 4200, "complexity_index": 187.3, "unmeasured": false},
    {"stage": "legacy-perl", "sloc": 0, "complexity_index": null, "unmeasured": true}
  ]
}
python3 -B "${CLAUDE_PLUGIN_ROOT}/scripts/befund_cli.py" resolve-output-path --repo <repo_root> --filename COMPLEXITY_SCORE.md
python3 -B "${CLAUDE_PLUGIN_ROOT}/scripts/befund_cli.py" resolve-output-path --repo <repo_root> --filename complexity_score_summary.json

Read-only constraint

Never use Write/Edit outside the resolved output paths.

04 同仓库更多 Skills