batuda

This skill should be used when the user asks to "debug", "diagnose", "check health", "run doctor", or says "app not working", "server down", "login broken", "blank page", "white screen", "500 error", "CORS error", "auth issue", "can't connect", "database error", "migration failed", "services not running", "certificate error", or asks to check if the Batuda server or web app is healthy locally.

This skill should be used when working with git worktrees for parallel Claude/dev sessions on Batuda — creating, provisioning, running, listing/diagnosing, or tearing down a worktree's dev stack, or cleaning up orphaned data. Covers the shared-services model (one Docker stack; each worktree gets its own Postgres database + MinIO bucket), the `pnpm cli worktree` commands, lifecycle hooks, and per-worktree CORS/portless.

Run the research quality eval (packages/research golden set) against the live pipeline and read the numbers. Use when asked to "run the eval", benchmark research grounding/recall, or take a before/after on a research change. Encodes the routing, cost, and safety guardrails so a run is not misconfigured.

Generates consistent git commit messages following project conventions. Use when committing changes, creating PRs, or when asked to write commit messages. Also validates if proposed messages follow project format.

This skill should be used when the user asks to "release", "deploy", "ship", "create a new version", "release server", "release internal", "release ui", "release mail-worker", "release all", or mentions CalVer versioning. Handles interactive release workflow for Batuda targets (server, internal, ui, mail-worker, or all) with independent CalVer versioning, GitHub tag-based deploys/publishes, and post-release verification.

Review the current diff for regressions, correctness bugs, tests, simplifications, and docs issues, scaling depth to a low/medium/high/xhigh/max effort level. Use when the user asks to review changes, review a diff/branch/PR, or runs /base-ui-review. Pass --comment to post a top-level PR comment, --comment inline for inline PR comments, or --fix to apply findings.

This skill should be used when the user asks to "open a PR", "create a pull request", "raise a PR", "PR this", or "send it for review" — turning a finished change (usually already committed on a local branch) into a reviewable, mergeable pull request. Drives the full Batuda PR lifecycle — re-baseline on main, verify end-to-end (with screenshots/recording for UI changes), self-review for pattern drift and AI-code smells, commit via /commit when there are new changes, run the verification gates, write an architecture-aware PR body in the house style, work the review loop (watch CI, address comments), and merge by rebase or squash. Stops at merge into `main`; deploying a merged change to production is `/release`.

This skill should be used when the user asks to "create an issue", "open an issue", "file an issue", "raise an issue", "log a bug", "track this as an issue", "make a GitHub issue", or describes a bug / feature / task to capture in the tracker. Interviews the user, researches the repo for real references and duplicates, drafts a type-aware (bug / feature / task) issue in Batuda's conventions, shows it for approval, then creates it via gh.

Audits a multi-day, multi-slice plan for big-picture coherence and silent drift. Spawns adversarial agents that argue against each other to surface defects that survive scrutiny — combats hallucinated findings (false positives) and missed issues (false negatives) by structured disagreement, not parallel agreement. Use between slices of a long plan, before starting the next slice, or when context has been compacted and the original purpose feels distant.

Explains how to develop Playwright - add APIs, MCP tools, CLI commands, and vendor dependencies.

Automate browser interactions, test web pages and work with Playwright tests.

Inspect Playwright trace files from the command line — list actions, view requests, console, errors, snapshots and screenshots.

This skill should be used when the user asks to "add a company", "update a company", "log an interaction", "create a document", "check the pipeline", "get next steps", "check overdue tasks", "create a page", "publish a page", "research a company", or mentions CRM data, MCP tools, companies, interactions, documents, tasks, or pages.

Drafts, reviews, and advises on sales-funnel copy (hero, CTA, landing page, about page, lead magnet, pricing page, email sequence, manifesto, webinar) using Russell Brunson's DotCom / Expert / Traffic Secrets frameworks — Hook-Story-Offer, Epiphany Bridge, Value Ladder, Application Funnel, Attractive Character, New Opportunity vs Improvement Offer, Big Domino, False Belief Patterns, Stack Slide, Hormozi Value Equation, Schwartz awareness levels, Soap Opera Sequence. Use when the user asks to write copy, draft a hero, rewrite the landing page, generate sales copy, write a CTA, write a lead magnet, write the about page, improve the homepage wording, fix the copy on X, draft an email sequence, OR asks a review/advise question about an existing artefact they present (a markdown doc, an i18n file, a pasted block of copy, a URL) such as "what would you change of X based on Y", "review this landing page", "is this hook strong", "which framework for...". Product-agnostic — loads project-specific context (services cat