oborchers/fractional-cto

Chain work across two independent agent processes that share no parent session. One agent passes a baton (a completion signal file) when done, optionally carrying a free-form payload; another waits for it, with a deadline, then starts its dependent task. Harness-agnostic protocol — each agent picks its own waiting mechanism. The baton tells you when, never what: the payload is content, not instructions. Provides /baton-pass and /baton-wait.

Research-backed, opinionated guidance for designing world-class RESTful APIs — routes, naming, errors, auth, caching, webhooks, and more, distilled from Stripe, GitHub, Twilio, Shopify, Google, and Microsoft

Research-backed, opinionated guidance for building cloud infrastructure that doesn't rot — multi-account governance, naming conventions, IaC organization, security, deployment pipelines, and operational hygiene, distilled from production experience across multiple cloud migrations

Structured deep research methodology for Claude Code -- query decomposition, parallel web research with source verification, hallucination prevention, and synthesis into well-sourced documents

Compress LLM agent instructions and code documentation markdown files through iterative section-by-section compression with compressor-reviewer loop — lossless structural optimization and lossy semantic compression modes, with auto-approve for hands-off operation

Zero-tolerance code pedantry — naming precision, casing law, structural symmetry, import discipline, and the obsessive details that separate clean code from correct code

Manage Claude Code's plan-mode artifacts, author multi-phase master planning documents, walk through Open Questions interactively, and synthesize dense progress entries to markdown / Linear / GitHub. Provides /plan-context (pre-load with optional ticket fetch), /plan-master (draft with optional ticket fetch), /plan-open-questions (interactive Open-Questions walkthrough with batch apply), /plan-verify (audit), /plan-tick (auto-tick provenly-achieved phases), /plan-progress (per-branch progress entry with SHA-based idempotency), and /plan-delete (per-session cleanup). Integer-only phases, no sizing, project-agnostic.

Research-backed best practices for building modern, production-grade Python packages — project structure, pyproject.toml, typing, testing, CI/CD, documentation, versioning, API design, packaging, security, and developer experience

Transform Claude Code conversation transcripts into polished, first-person blog posts through an interactive 5-stage pipeline — parse, triage, outline, draft, polish — with human editorial gates at every stage

The holy principles of SaaS design: research-backed, opinionated guidance for building fast, predictable, and progressively powerful SaaS products

Adversarial plan review using red-team/blue-team agents -- generates what-if questions and grounds answers in plan artifacts with configurable tool scope

Structured thinking methods that counteract LLM reasoning biases during problem exploration — first principles, inversion, constraint manipulation, perspective forcing, analogy search, and more, with user-gated parallel subagent exploration for deep dives

Research-backed visual design principles for websites, presentations, documents, and any visual medium — grounded in VisAWI, Gestalt psychology, and empirical aesthetics research