Modular suite for reproducible agentic AI workflows.

ai4X is a modular suite for reproducible agentic AI workflows. This repository is the integration, installation, and governance repository of that suite. The suite includes the root repository, the domain-owning modules ask, kob, ccp, and tcp, the shared module ccm, and the adjacent project-template repository ai4x-tpl.

ai4X is not only a modular suite for reproducible agentic AI workflows. It is also a reference implementation of a strict agent-first development model and an agent-guided documentation model for human-facing onboarding and support.

ai4X applies a strict agent-first development model.

Consequences:

• Deterministic checks are handled programmatically through TypeScript tests, shell checks, and deterministic validation commands such as verify and doctor.
• Higher-order semantic review, judgment, proposal assessment, handover, and capability assessment are intentionally designed for execution with agentic AI actors.
• In this model, developer and operator roles are primarily modeled as agentic AI roles.
• Therefore:
  • doc/usr/* is the direct human-facing documentation surface.
  • doc/agn/* is the agent-facing onboarding and navigation surface.
  • doc/arc/* is human-readable architecture and system-reference documentation.
  • adm/dev/* is agent-facing central ai4X development governance for suite-context work.
  • adm/ops/* is agent-facing central ai4X operations governance for suite-context work.
• The modules remain domain-owning repositories, but when they are used or evolved within ai4X, the canonical governance is owned here in the root ai4X repository.
• For contribution and governance entry, use ./CONTRIBUTING.md or hand ./doc/agn/maintainer-onboarding.md to your agent.

ai4X is, in that sense, built by agentic AI agents for agentic AI agents.¹

That development and documentation model is also why ai4X matters as more than a modular suite: it serves as a reference implementation for projects that want to apply the same approach in a governed and reproducible way.

In practice, that means:

• ai4X serves as a blueprint for projects that want to apply strict agent-first development in a governed and reproducible way.
• For new projects that want the same model in a minimal single-repo form, use ai4x-tpl, the ai4X-driven, minimal single-repo template for agent-first development and agent-guided documentation.
• ai4X separates governance, runtime orchestration, behavior curation, cognitive source-of-truth, technical source-of-truth, and cognitive materialization into explicit boundaries.
• ai4X distinguishes behavior, cognitive capabilities, and technical capabilities instead of collapsing them into one opaque prompt/tooling layer.
• ai4X treats agent-guided documentation for human-facing onboarding and support as a primary product surface rather than a README afterthought.
• ai4X includes Mechthild as a suite-integrated, need-first curator for high-quality project-local agents; see ./mod/kob/doc/usr/primer.md.
• ai4X gives those curated agents a flexible runtime surface through ask, which supports both public and project-local agents in one agent/runtime model.
• ai4X keeps cognitive bundle materialization deterministic through ccm as the single shared core used by curation and validation flows.
• ai4X applies explicit contracts, semantic review protocols, and deterministic validation gates to capability and suite evolution.
• ai4X deliberately uses a small, minimally invasive tool stack.

For Humans

Copy and paste this prompt to your LLM agent. Do not just read it yourself; let your agent explain ai4X to you:

I am completely new to ai4X.
Guide me through ai4X orientation, value, installation, and first hands-on use by following these sources in order:
1. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/doc/agn/user-onboarding.md
2. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/doc/usr/primer.md
3. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/INSTALL
4. https://raw.githubusercontent.com/normenmueller/ask/trunk/README.md
5. https://raw.githubusercontent.com/normenmueller/ask/trunk/doc/usr/primer.md
6. https://raw.githubusercontent.com/normenmueller/kob/trunk/README.md
7. https://raw.githubusercontent.com/normenmueller/kob/trunk/doc/usr/primer.md
8. https://raw.githubusercontent.com/normenmueller/kob/trunk/agn/curation/mechthild/doc/primer.md
Then:
- explain in beginner-friendly terms what ai4X is, what it is for, what is special about it, and what practical benefit I get from it
- explain the difference between public agents, project-local agents, `ask`, `kob`, and Mechthild without overloading me with governance detail
- propose one concrete first hands-on path for me; if I give no preference, default to the safest useful first path
- after that, ask me which path I want next: overview only, first public agent, project-local agent through Mechthild, or architecture/governance depth

For Agents

Use this bootstrap prompt when entering ai4X from a local checkout or from GitHub:

You are entering ai4X for governed work.
If you have a local checkout, use the local equivalents of these sources. Otherwise fetch them directly:
1. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/AGENTS.md
2. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/adm/dev/protocols/workflow.md
3. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/CONTRIBUTING.md
4. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/INSTALL
5. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/doc/agn/maintainer-onboarding.md
6. https://raw.githubusercontent.com/normenmueller/ai4x/trunk/doc/agn/source-map.md
Then summarize the suite structure, tell me which protocol applies to the current task, and do not infer workflow from repository layout alone.

License

See LICENSE. © 2026 nemron (#hgtt)

Footnotes

1. #eyodf = "eat your own dog food". ai4X applies its own agent-first development and operations model to itself. ↩