Generate a Paperclip company from a markdown brief.
Paperclip Blueprints is a CLI that turns a short brief (identity, north star, goals, constraints) into a complete, schema-validated Paperclip company bundle. Ready to import.
A management layer for an AI workforce.
Paperclip is an open-source platform for orchestrating multi-agent companies. It gives agents org charts, budgets, approval gates, and task tracking: the operating structure a real company runs on.
It's powerful, but a single company isn't one config file. It's dozens of markdown files covering identity, agent definitions, projects, operations rules, and governance. Writing all of that by hand is slow, and the result tends to drift.
Org charts
Roles, reporting lines, and escalation paths for every agent.
Task tracking
Projects and tasks the agents pick up, run, and report against.
Approval gates
Human-in-the-loop checkpoints before agents act on the big calls.
Budgets
Spend limits and governance, enforced as the company operates.
From a brief to a deployable bundle.
The right input isn't config. It's a brief. Describe what the company is, what it's emphatically not, the outcomes it chases, and the one north star everything serves. The structure follows from it.
A markdown brief: identity, "we are / we are not", goals as outcomes, north star.
Synthesizes the COMPANY.md identity, the hardest part to write by hand.
Derives agents, mandates, and span-of-control from canonical patterns.
Schema-checks every file against Paperclip's import format before you run it.
A brief, not config
You describe the company the way you'd describe it to a person. The org, roles, and rules are derived, not hand-typed.
Best practices by default
Every bundle encodes Paperclip's documented patterns, so they don't have to be remembered and re-applied each time.
- goal-as-outcome
- we are / we are not
- sensible span-of-control
- coherent governance
Schema-validated
Structural problems surface before the agents run on them. Goal-as-outcome is enforced at input time, catching the most common mistake.
What lands in the bundle.
A directory tree matching Paperclip's import format: identity, an auto-generated README with a mermaid org chart, operations rules, and a folder per agent with its role, mandate, and heartbeat.
All of it conforms to the paperclip/v1 and agentcompanies/v1 schemas.
On Paperclip & agent ops.
Field notes from running multi-agent companies in production.
Featured · PaperclipHow to use Paperclip Blueprints and wire the company it gives you
Paperclip Blueprints builds an agent company from a markdown brief. The import carries the operating structure: agents, adapters, models, budgets, projects, routines, skills. Here's what still needs your hands, and where the governing half of the brief has to go if agents are ever going to read it.
Read the article →
Agent SecurityAI agent sandboxing is the layer your fleet needs after the audit trail
AI agent sandboxing is the enforcement layer that sits after the audit trail. What run isolation actually means for a fleet operator and where the stack is going.
Read the article →
Agent SecurityAI agent credential security in production
Secrets leak across agent runs when every agent shares one credential pool. Vault-backed run access keeps secrets out of your AI agents.
Read the article →