Agent operating guide

Give agents stable routes, stable JSON, and exact next steps.

This page explains how agents should read LogBrew without needing browser state, hidden context, or private dashboard access.

Open docsRead Markdown
First safe read
logbrew status --json

Start with local status, then read logs, issues, actions, traces, releases, and projects only after auth and reachability are clear.

Entry points

Use predictable public URLs

Agents should find public LogBrew context through stable localized pages, Markdown mirrors, llms.txt, robots.txt, and sitemap.xml.

  • Use /en/agents.md when plain text is the safest input.
  • Use /en/docs.md for command and API route context.
  • Use /llms.txt to discover public agent-readable resources.

CLI recovery

Start with token-safe status

The CLI should tell agents whether local auth, API reachability, and recovery steps are ready without exposing token material.

  • Run logbrew status --json before project reads.
  • Follow next steps from JSON and human output instead of guessing.
  • Keep token material out of prompts, logs, screenshots, and public pages.

Read paths

Prefer resource-specific reads

Agents should read logs, issues, actions, traces, releases, and projects through the same resource vocabulary used by humans.

  • Use filters for release, environment, project, trace, level, status, name, and user.
  • Use issue and trace detail routes when an identifier is known.
  • Use Markdown mirrors for public context and the authenticated API for private project data.

Boundaries

Keep private data behind auth

Public agent surfaces must stay readable, while dashboard and project data remain protected by backend auth and noindex metadata.

  • Do not scrape dashboard data without a valid user session.
  • Do not publish private project names, tokens, traces, or deployment details.
  • Use public pages for general learning and authenticated routes for scoped user data.