Agent content packages

npm for
agents.

Package the knowledge, instructions, prompts, rules, and playbooks your agents need to know and do. Write in Markdown, publish to a registry, install onto agents and domains, and query at runtime.

See how it works View quickstart
# Pack, validate, and publish
npx pagenti init
npx pagenti pack . --dry-run
npx pagenti publish .

Built for developers shipping agentic systems.

content/ & execution/
@cognni/vendor-risk
agent runtime
// Querying installed execution playbooks
npx pagenti query \
--type execution \
--skill-code proceduralKnowledge \
--role ciso
// Returns:
✓ Execution: proceduralKnowledge:review-external-access

Your agents are running on scattered content.

Prompts live in code. Playbooks live in docs. Rules live in someone’s head. Knowledge and behavior drift apart.


Scattered Knowledge
Policies, Wikis, Notes

Prompt Sprawl
Hardcoded strings & text files

Runbooks
PDFs, internal portals
@cognni/vendor-risk
v1.2.0
Knowledge + Execution

Pagenti turns scattered agent content into installable packages.

Inside a package

What goes inside a Pagenti package?

Pagenti packages the full agent content stack: what the agent knows, how it acts, and how both are classified.

Knowledge

What the agent knows.

Expert context, guidance, facts, and critical questions the agent uses to reason.

Catalog: pagenti_knowledge
Crucial

Execution

How the agent acts.

Instructions, prompts, rules, procedures, and playbooks that drive the agent’s behavior.

Catalog: pagenti_execution

Skills

How content is classified.

The skillCode taxonomy that makes knowledge and execution searchable by type.

Catalog: pagenti_skills

A Pagenti package looks like an npm package.

But instead of code modules, it ships Markdown knowledge and execution files.

@cognni/vendor-risk/
├── pagenti.json           // Package manifest
├── pagenti-lock.json      // Dependency lockfile
├── content/
│   └── external-access-review.md   // Knowledge Markdown
├── execution/
│   └── review-external-access.md   // Instructions, prompts, rules
├── schemas/
└── assets/

{
  "name": "@cognni/vendor-risk",
  "version": "1.2.0",
  "knowledge": [
    {
      "code": "external-access-review",
      "path": "./content/external-access-review.md",
      "skillCode": "riskKnowledge",
      "role": ["ciso"]
    }
  ],
  "execution": [
    {
      "code": "review-external-access",
      "path": "./execution/review-external-access.md",
      "skillCode": "proceduralKnowledge",
      "role": ["securityAnalyst"]
    }
  ]
}

# External Access Review

## Purpose
Help a security analyst evaluate whether external access is justified.

## Critical Questions
- Who has access?
- Why do they need it?
- When should it end?

# Execution: Review External Access

## Instructions
1. Check the recipient domain against the allowed vendors list.
2. Require a documented internal data owner for the share.
3. Escalate unknown external access to the CISO immediately.
4. Auto-revoke access after 30 days unless exception is filed.

## Prompt
You are a security analyst evaluating external access. Apply the rules above...

Author. Publish. Install. Query.

npm ships code packages. Pagenti ships agent content packages.

1

Author

Create pagenti.json, then write knowledge in content/ and instructions, prompts, and rules in execution/.

npx pagenti init
npx pagenti pack . --dry-run # Validates paths and schema
2

Publish

Publish the package so other agents and teams can install the same versioned content.

npx pagenti publish .
3

Install onto an agent target

Install content onto an app, agent, and domain — so each agent gets the right knowledge, prompts, rules, and playbooks.

# Add package and map to target
npx pagenti add @cognni/vendor-risk@^1.2.0
# Or install globally to specific target
npx pagenti install @cognni/vendor-risk \
--app neo --agent neo --domain security
4

Query runtime content

pagenti search finds packages in the registry.
pagenti query retrieves installed knowledge or execution content for a running agent.

# Query installed content by context
npx pagenti query \
--type knowledge \
--skill-code riskKnowledge \
--role ciso

Find the right content for the right agent.

Every item can be labeled by role, domain, use case, tool, risk, regulation, and more — so agents retrieve only the content that fits the situation.

industry role domain useCase process tool risk regulation entityType contentType customerSegment maturityLevel

Not tags. Resolved context.

Before: "this is relevant for CISO"
After:
role pagenti_target_role/ciso
usageCount +1
scope inherited
indexed.role = ["ciso"]
External Access Review
role CISO
useCase Vendor Review
tool Entra ID
risk External Access

Agents query what they need.

Knowledge gives the agent context. Execution gives it instructions, prompts, and rules.

import { createPagentiRuntimeFromEnv } from "@x12i/pagenti-runtime";

const runtime = createPagentiRuntimeFromEnv();

// findKnowledge retrieves expert context to inform the agent
const items = await runtime.findKnowledge({
  skillCode: "riskKnowledge",
  role: "ciso",
  domain: "security",
});

// getExecution retrieves instructions, prompts, and rules ready for the agent loop
const playbook = await runtime.getExecution(
  "proceduralKnowledge",
  "review-external-access"
);

Registry data, powered by Catalox.

Pagenti is the package layer. Catalox is the governed data layer behind the registry.

1. Operators

Seed Catalox once. Requires @x12i/catalox@5.1.0.

2. Authors

Pack and publish validated packages.

3. Integrators

Install packages onto specific app/agent targets.

4. Agents

Query runtime content via SDK.

npm install @x12i/catalox@5.1.0
npx pagenti seed
// Seed initializes Catalox for Pagenti.
// Creates 23 catalogs, descriptors, and 20 SME skill codes.

A registry for reusable agent content.

Pagenti packages can be private, team-scoped, or public. Use them for internal agent capabilities, reusable industry packs, customer-specific content, or open-source playbooks.

@pagenti/base-security

v2.1.0 • Updated 2 days ago

Public

Core security principles, standard access review playbooks, and compliance baselines for general agent usage.

42 Knowledge 18 Execution

@cognni/vendor-risk

v1.2.0 • Updated 1 week ago

Private

Internal Cognni procedures for evaluating third-party vendors, data sharing agreements, and external access requests.

12 Knowledge 8 Execution

@x12i/agentic-execution

v3.0.0 • Updated 3 weeks ago

Public

Standardized execution instructions for tool-calling agents, reflection loops, and autonomous task planning.

5 Knowledge 24 Execution

@acme/sales-playbooks

v1.0.1 • Updated 1 month ago

Private

Objection handling, discovery guides, qualification flows, pricing logic, and vertical-specific messaging.

64 Knowledge 12 Execution

Start with the quickstart

Package manifest

Learn how pagenti.json maps content/ and execution/ files into installable agent content packages.

Smart properties

Learn how role, domain, use case, tool, risk, and skillCode help agents find the right content.

Runtime resolution

Query installed knowledge and execution content from your agent runtime.

Catalox backend

See how Catalox stores the registry, packages, versions, installs, and runtime content.

Package what your agents know and do.

Turn scattered prompts, rules, playbooks, and expert knowledge into installable agent packages.