brevio-core

Catches the mistakes coding agents repeat before they merge: a pull-request gate with no model call and no API key. scan, the agent hook and the MCP server use no network and send nothing.

In Plain English (For Anyone)	Technical Contract (For Engineers)
What it does: Checks code changes for common AI coding mistakes and repairs them locally before a pull request merges. When to use it: On your laptop with Claude Code or Cursor, or in GitHub Actions on every pull request.	CLI: brevio-core scan, brevio-core check, brevio-core task (connect, plan, build, qa, pr) Catalogue: 123 rules (13 scanned, 110 documented) Exit Codes: The exit code is 0 when nothing is found, 1 when not.

Track 1: Start & Single-Task Workflow

• Command reference: every command and option (scan, check, connect, plan, build, qa, pr, task, init --specs), generated from the code.
• The GitHub Action: inputs, permissions, events, merge queues and pinning to a commit.
• Questions people ask: why Python, why one line at a time, why an agent cannot silence it, how it compares.
• Changelog

Track 2: Rules & Auto-Fix

• Rules: one page per rule, saying what it catches, why, a bad and a good example, and how it is checked.
• Writing a rule: where a rule lives and what a pull request for one needs.

Track 3: Coding Agents & MCP

• Integrations: what each client gets and how to set it up.
• Agent tools: learn, hook install, gain, add, mcp and the agent skill.
• Claude Code, Cursor, Codex, Hermes, OpenClaw

Track 4: Formats & Guardrails

• Guards: .brevio/guards.json and the agent rules learn writes.
• SARIF: what scan --sarif writes for code scanning.

Track 5: Precision, AEO & Releases

• Precision on open-source repositories: sampled precision across 10 repositories and 19,126 files.
• Scoring a site for answer engines: brevio-core aeo, a separate command.
• Releases and listings: what a release does by itself, and the maintainers’ steps.
• Source and install instructions