perstack-ai · FL4TLiN3 · Feb 11, 2026 · Feb 11, 2026
diff --git a/docs/README.md b/docs/README.md
@@ -2,8 +2,6 @@
 title: "Perstack: The Agent Runtime"
 ---
 
-# Perstack: The Agent Runtime
-
 Define AI agents as declarative **Experts** in TOML. Execute them with deterministic, event-derived tracking. Each Expert runs in its own isolated context — no shared state, no prompt bloat, full execution history.
 
 - [Getting Started →](./getting-started.md)

diff --git a/docs/contributing/roadmap.md b/docs/contributing/roadmap.md
@@ -2,8 +2,6 @@
 title: "Roadmap"
 ---
 
-# Roadmap
-
 This roadmap outlines the direction of Perstack development. Priorities may shift based on community feedback and emerging requirements.
 
 > [!WARNING]

diff --git a/docs/getting-started.md b/docs/getting-started.md
@@ -2,8 +2,6 @@
 title: "Getting Started"
 ---
 
-# Getting Started
-
 This walkthrough takes you from zero to production integration. Each step addresses one or more of the [five structural problems](../README.md#why-perstack) in agentic app development.
 
 ## Prerequisites

diff --git a/docs/guides/README.md b/docs/guides/README.md
@@ -2,8 +2,6 @@
 title: "Working with Perstack"
 ---
 
-# Working with Perstack
-
 What are you trying to do?
 
 | I want to...                                         | Start here                                          |

diff --git a/docs/guides/adding-ai-to-your-app.md b/docs/guides/adding-ai-to-your-app.md
@@ -5,8 +5,6 @@ tags: ["integration", "container", "events"]
 order: 2
 ---
 
-# Adding AI to Your App
-
 You have an existing application — maybe a web app, a CLI tool, or a backend service. You want to add AI capabilities, but you're not sure where to start.
 
 **The challenge**: Agent frameworks often want to own the entire application. They're designed for building agent-first apps, not for embedding AI into existing systems.

diff --git a/docs/guides/extending-with-tools.md b/docs/guides/extending-with-tools.md
@@ -5,8 +5,6 @@ tags: ["tools", "mcp", "skills"]
 order: 3
 ---
 
-# Extending with Tools
-
 Your agent can reason and generate text. But sometimes that's not enough — you need it to search the web, query a database, or call an API.
 
 **The problem**: Integrating external tools usually means writing code, managing connections, and handling errors. Each new tool is a custom integration.

diff --git a/docs/guides/going-to-production.md b/docs/guides/going-to-production.md
@@ -5,8 +5,6 @@ tags: ["deployment", "container", "production"]
 order: 5
 ---
 
-# Going to Production
-
 Your agent works in development. Now you want to deploy it for real users.
 
 **The concern**: AI agents have larger attack surfaces than typical applications. They make decisions, use tools, and interact with external systems. How do you deploy them safely?

diff --git a/docs/guides/rapid-prototyping.md b/docs/guides/rapid-prototyping.md
@@ -5,8 +5,6 @@ tags: ["beginner", "quick-start", "create-expert"]
 order: 1
 ---
 
-# Prototyping for Agent-First Apps
-
 When you build an agent-powered app, the instinct is to start with the app — set up a project, install dependencies, write scaffolding. Then somewhere in the middle, you start figuring out what the agent should actually do.
 
 This is backwards.

diff --git a/docs/guides/taming-prompt-sprawl.md b/docs/guides/taming-prompt-sprawl.md
@@ -5,8 +5,6 @@ tags: ["prompts", "delegation", "architecture"]
 order: 4
 ---
 
-# Taming Prompt Sprawl
-
 Your agent started simple. A focused prompt, clear behavior, reliable results.
 
 Then requirements grew. You added more capabilities, more edge cases, more instructions. Now your prompt is 50,000 tokens long — a small novel — and the agent gets confused between tasks. You're afraid to change anything because you might break something else.

diff --git a/docs/making-experts/README.md b/docs/making-experts/README.md
@@ -2,8 +2,6 @@
 title: "Making Experts"
 ---
 
-# Making Experts
-
 Experts are defined in `perstack.toml`. This section covers how to define, test, and publish Experts.
 
 ## The essentials

diff --git a/docs/making-experts/base-skill.md b/docs/making-experts/base-skill.md
@@ -2,8 +2,6 @@
 title: "Base Skill"
 ---
 
-# Base Skill
-
 Every Expert automatically has access to `@perstack/base` — a built-in skill that provides file operations, runtime control, and other essential tools.
 
 **No configuration needed. It's always available.**

diff --git a/docs/making-experts/best-practices.md b/docs/making-experts/best-practices.md
@@ -2,8 +2,6 @@
 title: "Best Practices"
 ---
 
-# Best Practices
-
 These principles help you avoid common pitfalls in agent development: monoliths, complexity explosions, debugging nightmares, and fragile systems. Building a large agent head-on almost always fails.
 
 > [!NOTE]

diff --git a/docs/making-experts/examples.md b/docs/making-experts/examples.md
@@ -2,8 +2,6 @@
 title: "Examples"
 ---
 
-# Examples
-
 Patterns for defining Experts. Each example highlights a specific skill type or integration approach.
 
 - [GitHub Issue Bot](#github-issue-bot)

diff --git a/docs/making-experts/publishing.md b/docs/making-experts/publishing.md
@@ -2,8 +2,6 @@
 title: "Publishing Experts"
 ---
 
-# Publishing Experts
-
 Share your Experts with the community by publishing them to the Perstack Registry.
 
 ## Before publishing

diff --git a/docs/making-experts/skills.md b/docs/making-experts/skills.md
@@ -2,8 +2,6 @@
 title: "Skills"
 ---
 
-# Skills
-
 Skills give Experts access to external tools via MCP (Model Context Protocol).
 
 | Type               | Use case                                             |

diff --git a/docs/making-experts/testing.md b/docs/making-experts/testing.md
@@ -2,8 +2,6 @@
 title: "Testing Experts"
 ---
 
-# Testing Experts
-
 Experts are probabilistic — the same query can produce different results. This guide covers strategies for testing effectively.
 
 ## Local testing

diff --git a/docs/operating-experts/deployment.md b/docs/operating-experts/deployment.md
@@ -2,8 +2,6 @@
 title: "Deployment"
 ---
 
-# Deployment
-
 Perstack runs on any Node.js environment. Choose your deployment target based on security requirements, cost, and operational preferences.
 
 ## Docker

diff --git a/docs/operating-experts/isolation-by-design.md b/docs/operating-experts/isolation-by-design.md
@@ -2,8 +2,6 @@
 title: "Isolation by Design"
 ---
 
-# Isolation by Design
-
 Perstack is designed to run inside isolated environments. The runtime itself doesn't enforce security boundaries — your infrastructure does.
 
 For the rationale behind this approach, see [Sandbox Integration](../understanding-perstack/sandbox-integration.md).

diff --git a/docs/operating-experts/observing.md b/docs/operating-experts/observing.md
@@ -2,8 +2,6 @@
 title: "Observing"
 ---
 
-# Observing
-
 Every Perstack execution produces a complete audit trail. Use events and checkpoints to monitor, debug, and verify Expert behavior.
 
 ## Event stream

diff --git a/docs/operating-experts/skill-management.md b/docs/operating-experts/skill-management.md
@@ -2,8 +2,6 @@
 title: "Skill Management"
 ---
 
-# Skill Management
-
 The Perstack runtime manages three types of skills through specialized Skill Managers. Each type has different initialization behavior and security implications.
 
 ## Skill Types

diff --git a/docs/references/cli.md b/docs/references/cli.md
@@ -2,8 +2,6 @@
 title: "CLI Reference"
 ---
 
-# CLI Reference
-
 ## Running Experts
 
 ### `perstack start`

diff --git a/docs/references/events.md b/docs/references/events.md
@@ -2,8 +2,6 @@
 title: "Events Reference"
 ---
 
-# Events Reference
-
 Perstack emits events during Expert execution for observability and state management. This document explains the event system architecture and how to consume events.
 
 ## Event Types Overview

diff --git a/docs/references/perstack-toml.md b/docs/references/perstack-toml.md
@@ -2,8 +2,6 @@
 title: "perstack.toml Reference"
 ---
 
-# perstack.toml Reference
-
 - [Complete Example](#complete-example)
 - [Runtime Configuration](#runtime-configuration)
 - [Expert Definition](#expert-definition)

diff --git a/docs/references/providers-and-models.md b/docs/references/providers-and-models.md
@@ -2,8 +2,6 @@
 title: "Providers and Models"
 ---
 
-# Providers and Models
-
 Perstack supports multiple LLM providers. Configure via CLI options, environment variables, or `perstack.toml`.
 
 ## Default Model

diff --git a/docs/understanding-perstack/boundary-model.md b/docs/understanding-perstack/boundary-model.md
@@ -2,8 +2,6 @@
 title: "The Boundary Model"
 ---
 
-# The Boundary Model
-
 [Isolation](./concept.md#isolation) is one of Perstack's three core goals. Experts are isolated from everything except their role — model access, context, tools, and dependencies are all mediated by the runtime.
 
 The **boundary model** extends this isolation to the application architecture level: **separate the application from the agent runtime**.

diff --git a/docs/understanding-perstack/concept.md b/docs/understanding-perstack/concept.md
@@ -2,8 +2,6 @@
 title: "Concept"
 ---
 
-# Concept
-
 Perstack is built on a concept called **Expert Stack** — the architecture that enables agent-first development.
 
 > [!NOTE]

diff --git a/docs/understanding-perstack/experts.md b/docs/understanding-perstack/experts.md
@@ -2,8 +2,6 @@
 title: "Experts"
 ---
 
-# Experts
-
 Experts are the core building block of Perstack — modular micro-agents designed for reuse.
 
 > [!NOTE]

diff --git a/docs/understanding-perstack/registry.md b/docs/understanding-perstack/registry.md
@@ -2,8 +2,6 @@
 title: "Registry"
 ---
 
-# Registry
-
 The Perstack Registry is a community-driven ecosystem for sharing Experts — like npm, but for AI agents.
 
 ## Definition-only

diff --git a/docs/understanding-perstack/runtime.md b/docs/understanding-perstack/runtime.md
@@ -2,8 +2,6 @@
 title: "Runtime"
 ---
 
-# Runtime
-
 The Perstack runtime combines probabilistic LLM reasoning with deterministic state management — making agent execution predictable, reproducible, and auditable.
 
 ## Execution model

diff --git a/docs/understanding-perstack/sandbox-integration.md b/docs/understanding-perstack/sandbox-integration.md
@@ -2,8 +2,6 @@
 title: "Sandbox Integration"
 ---
 
-# Sandbox Integration
-
 ## Why sandbox-first?
 
 AI agents differ fundamentally from traditional software. The same input can produce different outputs. Model updates can change behavior. Hallucinations can trigger destructive actions without any attacker involved.

diff --git a/docs/using-experts/README.md b/docs/using-experts/README.md
@@ -2,8 +2,6 @@
 title: "Using Experts"
 ---
 
-# Using Experts
-
 This section covers **running and integrating Experts** — CLI usage, state management, and error handling.
 
 ## In this section

diff --git a/docs/using-experts/error-handling.md b/docs/using-experts/error-handling.md
@@ -2,8 +2,6 @@
 title: "Error Handling"
 ---
 
-# Error Handling
-
 Perstack is designed to recover from errors automatically when possible. Most errors are fed back to the LLM for self-correction rather than crashing the run.
 
 ## How errors are handled

diff --git a/docs/using-experts/running-experts.md b/docs/using-experts/running-experts.md
@@ -2,8 +2,6 @@
 title: "Running Experts"
 ---
 
-# Running Experts
-
 Perstack provides two commands for running Experts: `start` and `run`.
 
 ## `start` vs `run`

diff --git a/docs/using-experts/state-management.md b/docs/using-experts/state-management.md
@@ -2,8 +2,6 @@
 title: "State Management"
 ---
 
-# State Management
-
 Perstack automatically saves execution state, enabling continuation and replay.
 
 ## Execution hierarchy

diff --git a/docs/using-experts/workspace.md b/docs/using-experts/workspace.md
@@ -2,8 +2,6 @@
 title: "Workspace"
 ---
 
-# Workspace
-
 The workspace is the isolated directory where Experts execute. All file operations are confined to this directory.
 
 ## Default workspace