Field guide / sub-domain
33
sources in this field
Updated September 10, 2026
Current thesis
The shortest path to orientation.
This synthesis records claims and practices from the cited sources; reported outcomes and product capabilities have not been independently verified.
Claude configuration operates at three layers: environment (--dangerously-skip-permissions for autonomy, autosave, recovery flags), context architecture (CLAUDE.md with declarative behaviors, .claude/rules/ YAML, _MANIFEST.md priority loading), and skill design (progressive disclosure, state-machine optimization). Tier permissions into Safe, Ask-first, and Human-owned categories. CLAUDE.md functions as a 'front desk, not a filing cabinet'—global CLAUDE.md holds universal rules while local/project CLAUDE.md holds project specifics; separate them to avoid bloat. CLAUDE.md (auto-loaded, under 200 lines) stays distinct from brain.md (ICP/personas/scoring, referenced via @brain.md) so GTM substance doesn't bloat unrelated runs. brain.md should end with explicit source-ranking (closed-won CRM > brain.md > old notes) preventing silent contradictions. MCP gotchas: unquoted env vars leak API keys into .mcp.json; explicit "type": "stdio" required; .mcp.json read only at session start. Never @-reference .env files; use settings.json deny rules and pass stripped .env.example instead. Use three-file hyper-personalization pattern: SOUL.md (voice/values, under 200 lines), USER.md (~4000 words), AGENTS.md (operational rules). Opus 4.8 upgrades at 4.7 pricing; /fast mode costs 3x less at 2.5x speed.
Evidence board
Claims worth carrying forward01Restructuring GTM around Claude Code as orchestrator (with tools like ColdIQ MCP as endpoints) cut a campaign build from two days of manual CSV-shuffling across four SaaS tools to one prompt and ~20 minutes of agent work.
02Project structure convention: CLAUDE.md (auto-loaded, under 200 lines, standing rules) is kept separate from brain.md (ICP/personas/scoring, referenced explicitly via @brain.md) so GTM substance doesn't bloat context on unrelated runs.
03brain.md should end with an explicit source-ranking block (e.g., closed-won CRM data > brain.md > last quarter's notes) so the model doesn't silently pick a winner when the ICP doc, intelligence store, and old notes contradict each other.
04MCP setup gotchas: unquoted env vars in `claude mcp add` cause shell expansion that leaks the live API key into .mcp.json; always write explicit "type": "stdio" or a url-only remote entry silently fails; .mcp.json is read only at session start.
05Never let Claude @-reference .env files—contents flow into the prompt, session transcript, and downstream outputs with no default block. Use a settings.json deny rule (Read(./.env) etc.) and hand the model a stripped .env.example instead.
06Seven-step outbound loop: detect signal → score/tier (1-100 via scoring.md) → find contacts → enrich/validate (email then phone, phone ~10x cost of email) → generate tiered copy → route to sequencer → analyze results back into the intelligence store.
07Skills files (.claude/skills/*/SKILL.md) encode one procedure each (e.g., cold-email-copy) so repeated prompts produce consistent output instead of drifting each time it's typed differently; Claude loads only relevant skills, keeping context small.
08Scraping LinkedIn profiles breaches its User Agreement and hiQ v. LinkedIn ended in a $500,000 judgment against the scraper with a data-deletion order; job-post and tech-stack signals can be sourced legally instead via Greenhouse/Lever APIs and BuiltWith/Wappalyzer.
Adjacent fields
Latest evidence
Michel Lieben
A GTM agency's detailed playbook for running outbound campaigns through Claude Code as the orchestrator, with Apollo/Clay/Instantly-equivalent tools (via MCP) as endpoints. Covers folder structure, MCP connection setup, a seven-step signal-to-meeting workflow, skills files, reusable prompts, and eight recurring failure modes.
Charlie Hills
Anthropic's Claude Code team formally defines agentic 'loops' as cycles of work repeated until a stop condition is met, categorizing them into four types (turn-based, goal-based, time-based, proactive) by what gets handed off. Charlie Hills synthesizes the framework with practical commands (/goal, /loop, /schedule, skills) and a warning that every loop still needs an explicit stop condition or it burns tokens unattended.
Harshil Tomar
A curated set of Claude Code workflow tricks—sourced from Matt Pacock's skills, Emil Kowalski's micro tips, and personal practice—focused on plan interrogation, TDD enforcement, git safety, terminology consistency, and converting chat into scoped work.
The Startup Ideas Podcast (SIP) 🧃
A structured framework for using Claude Code as an 'AI employee' rather than a chatbot, covering repo-as-workspace, memory files (CLAUDE.md/roadmap.md/review.md), plan-mode briefs, scoped tickets, visual verification, three-layer review, scheduled routines, parallel agents, tiered permissions, and reusable skills/hooks — demonstrated via a med-spa missed-lead responder build.
Matt Pocock
Matt Pocock shares an experimental global CLAUDE.md instruction: always write in ASD-STE100 Simplified Technical English, and always read CONTEXT.md files to adopt their ubiquitous language.
Lydia Hallie ✨
Claude Code supports user-defined output styles via config files, letting users control response tone/format (e.g., a casual style for end-of-day use), configured through /config → Output style.
CyrilXBT
A detailed 6-step system for building an Obsidian vault that auto-captures content via Readwise, Airr, Whisper, and a Telegram bot, routes everything through N8N pipelines, and uses Claude to generate daily briefings and weekly syntheses. The core thesis: knowledge systems fail because they optimize for input, not output—feedback loops (daily briefs, weekly synthesis) are what turn a dead archive into a thinking partner.
Charlie Hills
gstack is Garry Tan's open-source Claude Code setup that transforms a solo AI assistant into a structured 6-role virtual engineering team (CEO, Eng Manager, Designer, QA, Release Engineer, Doc Engineer) via 23 slash-command skills. Tan claims ~810× his 2013 productivity rate measured in normalized logical lines of code, shipping 3 production services and 40+ features in 60 days part-time. The system enforces a think→plan→build→review→test→ship→reflect sprint discipline where each skill feeds context into the next, and supports 10-15 parallel sprints via Conductor.
Ruari Wood-Freeman
A GitHub-based plugin marketplace pattern allows centralized management of multiple Claude plugins with sub-skills and agents across an entire organization. Updates to the central repo automatically sync to all users' Claude instances, eliminating manual distribution overhead.
Aakash Gupta
Hannah Stulberg (DoorDash PM) built a shared git repo where every team function checks in context—call summaries, decision logs, analytics queries—queryable in natural language via Claude Code. Four independent implementations (DoorDash, Pendo, Google, solo) converged on the same 3-layer architecture, suggesting a generalizable pattern for solving institutional knowledge loss.