Codebase Argusv0.1.0

Platform engineering teams can wire Codebase Argus into GitHub via the webhook server so every opened, reopened, or ready_for_review PR receives an evidence-first review with inline comments. The tribunal mode combines OpenAI, Claude, and Codex findings to reduce single-model blind spots. Teams stay in control because Argus posts COMMENT-only reviews and requires explicit approval before merge.

Developers paste failing GitHub Actions logs or let Argus fetch job logs directly with a read-only token to get root-cause analysis on the first failing command. The tool suggests the smallest verifiable fix, cutting mean time to recovery for red builds. This is especially useful for monorepos where multiple workflows fail simultaneously.

Open-source maintainers and enterprises running long-lived forks can use the downstream merge/rebase workflow to periodically sync against the upstream repository. Argus reviews the integration work the same way it reviews PRs, flagging conflicts in workflows, lockfiles, and route handlers. This keeps forks current without manual diff audits.

Security and compliance teams use Argus to flag PRs touching auth, tokens, webhooks, payments, signatures, and route handling, plus workflow edits like pull_request_target. Because reviews are evidence-first and read-only, findings can be routed into a compliance trail. The tool never approves or pushes, aligning with change-management controls.

Development consultancies can offer 'AI-assisted code audits' as a productized service, using tribunal mode to generate multi-provider review reports for client repositories. Cloning the checkout and running `npm run argus -- review` against client PRs produces shareable markdown reports. The agent-first playbook also works as Claude Code or Codex project instructions for on-site engagements.

The CLI and playbook remain free and portable for individual developers, while the hosted Next.js webhook server (App manifest, inline comments, PR commands) is a paid cloud subscription per installation. Revenue comes from teams wanting zero-setup automatic review on every PR.

The product itself is free to self-host, but organizations pay for advanced policy packs, tribunal orchestration, audit logging, and SSO. Provider costs stay with the customer via their own OpenAI, Anthropic, or Gemini API keys, lowering the vendor's COGS.

Consultancies and MSPs resell Argus-powered review retainers, bundling the tool with human engineers who triage findings and implement autofix plans. The open-source core keeps delivery costs low while retainer fees capture ongoing value.