How to use HQ

HQ Core v15 is live. The biggest release yet. HQ is a shared context layer that syncs your knowledge, skills, policies, and capabilities across every agent you run. v15 makes that real in two ways: it goes where your agents go, and it works the same wherever you run it. — 🧩 HQ now runs inside Cowork New hq-pack-cowork puts native HQ inside Cowork — and any Claude Code plugin host — through a host-launched MCP server wrapping the real hq qmd binaries (20 tools) plus 10 skills. A sandboxed agent reaches your host auth and runs the actual CLI, so your whole HQ — vault, search, secrets, sync — comes with it. Hardened before merge: two security audits, a code review, and a live MCP smoke test. Read-only CLI allowlist, secret tools that refuse to launch a shell, first-party-only installs. — 🔁 The same HQ, whichever agent you run Full Codex permission hook parity with Claude Code. Same skills, same guardrails, same policies on Claude Code, Cursor, and Codex. The cross-agent operating system, actually delivered end to end. — 👥 Built for the whole team, not just coders • Quiet, plain-language voice by default — HQ speaks in outcomes, not jargon (/output-style hq-operator flips on the full technical play-by-play) • /dm — message a teammate from a session; lands in their HQ Sync menubar. Attach agent context for a one-click "Copy prompt," or schedule it • /work-broadcast — right-sized Slack announcements that post as you • /ontology — situational awareness per company: what's been decided, what's hot, who's active • Company-scoped writes verify the company first — no cross-tenant leaks • Bind a company mid-session and its hard policies surface automatically, with a just-in-time reminder before any infra/credential command — ✨ More new capabilities • /humanize — strips the AI tells out of anything HQ writes; runs automatically before finished prose • /hq-heal — paste a Claude/Codex error mid-session, get a diagnosis fix, auto-files a bug • /hq-share now drafts the share note for you • hq-files cloud retrieval — browse/cat/get from the vault without syncing the whole thing — 🛡️ Upgrades that never cost you anything • /learn now survives upgrades (routes to personal/, never wiped core) • Customizations live in personal/ or companies/, never in the wiped-on-upgrade core • Capability links are mechanically enforced as one-time, never logged • AI-velocity time sense — estimates in agent-sessions, not human weeks — ⚙️ Under the hood • Engineering is now an installable pack — 17 dev skills (/tdd, /review, /execute-task, /run-project, /architect, /diagnose, /land…), 6 workers, knowledge policies moved out of core into hq-pack-engineering. Founders/ops/sales start lean; coders add it with one command; upgraders get it auto-installed and symlinked back — zero prompts, nothing lost • Opus 4.8 plan mode as the defaults — HQ proposes before it acts • Ralph mode re-engined — unattended in-session loop, no per-story subprocess, same on Claude Code and Codex • Live session titles — sidebar/tab/resume show company · project · command • Windows compatibility — hq install and the installer run clean under Git Bash • Search reindexes automatically after every sync — The thesis: HQ goes where your agents go. The same HQ, whichever agent you run. Lean by default, powerful on demand. Upgrades that never cost you anything. Full notes ↓ github.com/indigoai-us/hq-co…

/deploy is live! HQ Deploy allows any HQ user to ship fully functional web applications with a single command Any idea, dashboard or deck can now be an app Goodbye Vercel, Goodbye Netlify

HQ Core v14 is live. /deploy — ship decks, reports, dashboards, and web builds to live URLs in seconds, with auto-protection for sensitive data hq-secrets — secret injection without leaking values to stdout HQ Cloud — synced, cloud-backed companies from the CLI Identity layer — login/logout/whoami with Cognito auth /import-claude — import Claude sessions, MCPs, skills, hooks, repos, and knowledge into HQ /hq-files — company-level file access controls New quality gates for AI coding: /architect /review-plan /review /diagnose /out-of-scope /adr Estimate calibration: Track estimates vs actuals over time so agents can improve, not just guess. Per-company audit memory: Every session is hardlinked into the company it touched with a persistent audit trail. Agents don’t start cold anymore. Codex parity: 50/50 skills and 41/41 commands now paired across Claude Code and OpenAI Codex. New workflow primitives: • /discover • /land-batch • /deep-plan Plus conflict resolution, ASCII graphics, a new Cavebro output style, and a full rework of companies/_template/. The thesis: Agents that remember. Agents that ship sharper. Agents that work the same way regardless of runtime. Full notes: github.com/indigoai-us/hq-co…

Introducing HQ: the AI operating system for your company. HQ lets your team share knowledge, skills, workflows and API access across the organization, so everyone works from the same foundation. Team AI is still messy, and hard. Individuals work in silos. There is no consistent knowledge base. Teams end up with 10 versions of the same skill, all performing at different levels. One person becomes 10x more effective, while everyone else is left trying to catch up. HQ makes this simple. Built on top of Codex and Claude Code, HQ sits directly inside your existing workflow. It helps manage your company's context layer, share skills, standardize workflows and control access, so people only see what they need to see. We built HQ internally to solve a real problem. We use it every day. It has been incredible for enabling our power users, but just as importantly, it has helped the least technical members of our team become incredibly effective AI users. HQ turns one person's breakthrough into everyone's baseline. It is a powerful, elegant tool, and we are incredibly excited to share it with the public. Book a demo below.