agentleFS
Yesterday-AIYesterday-AI/paperclip-plugin-company-wizard/AGENTS.md

paperclip-plugin-company-wizard

AGENTS.md171 starsChanged 8 months agoMITView on GitHub

Company Wizard is a Paperclip plugin for bootstrapping agent company workspaces from modular, composable templates. It provides an interactive wizard (manual and AI-powered paths) that assembles ready-to-run companies — roles, workflows, skills, process docs, heartbeat sections, and Paperclip API provisioning — all from a single UI. The plugin runs inside the Paperclip board UI and ships with a complete template library: 15 presets, 26 modules, 17 optional roles (CEO is the only base role). The default Paperclip company setup is…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository5 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every company starts functional and gets better as you add roles.

    line 30

≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

Company Wizard — Project Vision

What

Company Wizard is a Paperclip plugin for bootstrapping agent company workspaces from modular, composable templates. It provides an interactive wizard (manual and AI-powered paths) that assembles ready-to-run companies — roles, workflows, skills, process docs, heartbeat sections, and Paperclip API provisioning — all from a single UI.

The plugin runs inside the Paperclip board UI and ships with a complete template library: 15 presets, 26 modules, 17 optional roles (CEO is the only base role).

Why

The default Paperclip company setup is a blank slate: one CEO with a bootstrap prompt, no defined workflows, no process for generating issues, no review cycles. Every company starts from zero and reinvents the same patterns.

Company Wizard encodes organizational knowledge into reusable templates. Instead of hoping agents figure out how to collaborate, you start with proven structures — issue generation from roadmaps, auto-assignment of idle agents, stall detection, PR review flows — all wired up and ready.

Core Idea: Gracefully Optimistic Architecture

Inspired by the OpenClaw gateway architecture where channel adapters declare what they CAN do rather than what they MUST do, and the system degrades gracefully when features are absent.

Company Wizard applies the same principle to organizational capabilities: the system never asks "which preset is this?" — it asks "which roles are present, and what can they do?"

OpenClaw PatternCompany Wizard Equivalent
Channel declares capabilitiesModule declares capabilities with owners[] chain
Missing feature → graceful degradeMissing role → fallback owner takes over
Core is channel-agnosticAssembly is preset-agnostic
Adapter is optionalRole is optional — base always works

A company with just the CEO works fine — they handle backlog, auto-assign, and stall detection. Add an Engineer and they take over implementation. Add a Product Owner, and it automatically takes over backlog management as primary owner while the CEO becomes the fallback safety net. Add a Code Reviewer, and PR review workflows activate.

Every company starts functional and gets better as you add roles. No capability is ever "missing" — there's always someone responsible.

Design Principles

  • Files, not config servers — Company structure is markdown files on disk. Agents read them fresh every heartbeat. Edit a file, behavior changes next cycle.
  • Composable, not monolithic — Modules are independent building blocks. Presets are just curated module combinations. Everything can be mixed, matched, and extended.
  • Opinionated defaults, easy overrides — Templates encode best practices but every file is editable after generation (including in the wizard's preview step). Company Wizard gets you started; you own the result.
  • Capability-based, not identity-based — The system resolves "what can this company do?" based on present roles, not "which template was selected?" Roles declare capabilities, modules declare ownership chains, the assembly resolves at build time.
  • Primary/fallback ownership — Every capability has an ownership chain. The most qualified present role owns it; less specialized roles serve as safety nets.
  • Shared skills, role-specific overrides — Primary skills live in a shared skills/ folder unless a role brings a genuinely different approach. Fallbacks are always role-specific.

Architecture

src/
├── worker.ts                  # Plugin worker (actions: preview-files, start-provision, check-auth, ai-chat, check-ai-config, refresh-templates)
├── manifest.ts                # Plugin manifest (id, displayName, sidebar slot)
├── logic/                     # Pure functions (assembly, resolution, template loading, AI wizard)
├── api/                       # Paperclip REST API client + provisioning
└── ui/
    ├── context/               # WizardContext — state machine + reducer
    └── components/            # WizardShell, step components, ConfigReview (file preview)

templates/
├── roles/                     # All roles with role.meta.json
├── modules/                   # Composable capabilities (26 modules)
│   └── <module>/
│       ├── module.meta.json   # capabilities[], activatesWithRoles[], tasks[], goal?, adapterOverrides?
│       ├── skills/            # Shared primary skills
│       ├── agents/<role>/     # Role-specific overrides, fallbacks, heartbeat sections
│       └── docs/              # Shared docs injected into all agents
└── presets/                   # Curated combinations (15 presets, may include goals[])

Each module contains:

  • module.meta.json — Capability ownership chains, activation rules, initial tasks, inline goal
  • skills/<skill>.md — Shared primary skill (used by any primary owner)
  • agents/<role>/skills/ — Role-specific overrides and fallback variants
  • docs/ — Shared documentation injected into all agents

Where This Is Going

See ROADMAP.md for the full backlog.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.