@rulebased/core

Core logic for harness engineering auditing, recommendations, and initialization.

Part of @rulebased/harness.

API

import { audit, formatReport, formatScore } from "@rulebased/core/auditor";
import { recommend, formatRecommendations } from "@rulebased/core/recommender";
import { initHarness } from "@rulebased/core/initializer";
import { parseTranscript, computeStats } from "@rulebased/core/transcript";
import { evaluateLog, formatLogEval } from "@rulebased/core/log-evaluator";

audit(projectPath)

Runs 34 checks based on OpenAI Codex harness standards and returns a scored report.

recommend(report)

Generates prioritized recommendations from an audit report.

initHarness(projectPath, options?)

Creates AGENTS.md, CLAUDE.md, specs/, tasks/, .harness.json in the target project.

parseTranscript(filePath) / evaluateLog(stats)

Parses Claude Code JSONL transcripts and evaluates session compliance.

License

MIT