Can we please get a HOW TO GUIDE or more details about the OpenClaw runtime in the page?

Feature Requests
1

Can we please get a HOW TO GUIDE or more details about the OpenClaw runtime in the Pickaxe Learn page?
I am migrating my studio over to the OC runtime and would like to understand what is happening under the hood a little so I can apply some of what I have learned using OC.
Trying to figure out how Pickaxe and OpenClaw correlate.

Here’s how Cursor explained it to me:
Here’s a clean way to map what you know to Pickaxe, without assuming Pickaxe exposes the same filenames in the UI.

Two different “OpenClaw” surfaces

The stack that talks about SOUL.md, IDENTITY.md, TOOLS.md, etc. is usually the standalone OpenClaw workspace (local/project files, fixed load order). That’s a multi-file, filesystem-first design.

Pickaxe’s OpenClaw (see OpenClaw AI Engine | Pickaxe Learn) is Pickaxe-hosted: workspace toggle, per-user isolated runtime, optional upload/download of an OpenClaw configuration. Public Learn describes that at a high level; it does not spell out “your Role field = SOUL.md” in the same way the Claw workspace docs do.

So: treat the mapping below as conceptual (how to think and port content). Confirm the exact upload format by exporting a config from Pickaxe after you enable OpenClaw, or by inspecting whatever bundle Pickaxe accepts when you upload.

Conceptual map: Claw-style files → Pickaxe

OpenClaw-style file (typical role) Closest Pickaxe equivalent
SOUL.md — personality, values, voice, boundaries Mostly Prompt → Role (your *-toon.txt). Sometimes split: stable “soul” in Role + repeating rules in Model Reminder.
IDENTITY.md — name, how it introduces itself, presentation Intro message / ice breakers in the Prompt tab; first lines of Role (“You are Mindset Coach (Coach Lori)…”); agent title in Studio.
TOOLS.md — when to use which tools, tool discipline Actions tab (each action’s trigger text) + short Role bullets (“use Search when…”, “use Email User when…”). OpenClaw built-in tools (browse, PDF, code, computer link) are platform capabilities; you reflect policy in Role (what to prefer, what not to promise).
BOOTSTRAP.md — startup behavior, first-run Intro message, optional StartPoint onboarding; anything that runs once or sets expectations at session start.
USER.md — preferences, user-specific context User Memories in Studio + what OpenClaw persists per user (Learn: “deeper memory”). Your repo’s User Memories/ markdown is documentation / export, not the live store.
AGENTS.md — other agents, routing, delegation In Pickaxe: linked Pickaxes / handoff Actions + hub prompt (e.g. Manna Guide lists specialists). Your canonical suite map is docs/jp-manna-suite-vision-and-architecture-v26.4.md, not a single Pickaxe file named AGENTS.md.

How this helps you transfer OpenClaw knowledge

  1. Merge split files into layers Pickaxe actually has
    Often: SOUL.md + big chunks of IDENTITY.md + routing snippets from AGENTS.md → one Role TOON (plus Model Reminder for “every message” rules).

  2. Put tool policy in Role; put tool wiring in Actions
    TOOLS.md-style “when to call X” → Role + per-action triggers. Don’t duplicate long mechanics in both unless you need to.

  3. Treat USER.md as “memory design,” then implement in Studio
    Define keys and meaning in your docs; populate User Memories and agent behavior in prompts.

  4. Validate with Pickaxe export
    After OpenClaw is on, download the workspace/agent config Pickaxe offers and see whether it’s a zip with familiar .md names or something else. That answers “is it literally SOUL.md here?” for your account.