Campfire

Product Introduction

1. Definition: Campfire is an open-source, harness-independent workspace platform designed for human-AI collaboration in software development. It functions as a persistent, shared state layer that sits between developers and their existing AI coding agents (like Codex, Claude Code, or custom agents).
2. Core Value Proposition: Campfire exists to solve the fragmentation of work and context when multiple humans and AI agents collaborate. Its primary value is providing a unified, persistent workspace where goals, code changes, agent findings, and human decisions are recorded and shared, eliminating the "session amnesia" typical of isolated AI agent interactions.

Main Features

1. Persistent Workspaces: A Campfire workspace is a dedicated, versioned environment that persists beyond individual terminal sessions. It aggregates all activity—goals, code commits, agent-generated findings, and human decisions—into a single chronological timeline. Technically, it acts as a centralized event log and state manager for multi-participant (human and AI) workflows.
2. Goal-Oriented Collaboration: The workflow is initiated by defining a clear objective (e.g., "Ship the new billing flow"). This goal anchors all subsequent agent and human actions within the workspace, providing context and direction without requiring complex orchestration graphs or predefined agent chains.
3. Capability-Based Agent Security: Campfire implements a granular permissions model for AI agents. Instead of granting broad repository access, users define explicit capabilities per workspace: which directories an agent can read/write, which commands it can execute (e.g., run tests), and what it is explicitly denied (e.g., production credentials). This enables useful access with clear security boundaries.
4. Immutable Activity History & Audit Trail: Every action within a workspace is recorded with attribution (which human or agent performed it) and context. This creates a searchable, linear history of how the work evolved, which decisions were made, and why the final codebase looks the way it does, crucial for debugging and team onboarding.
5. Agent-Agnostic Integration: Campfire does not replace or orchestrate agents. It is designed to integrate with existing AI coding tools and custom agents. Agents interact with the workspace via its API or CLI, reading context from it and writing their outputs (code changes, findings) back to it, allowing heterogeneous agents to contribute to a shared goal.

Problems Solved

1. Pain Point: Context and Work Fragmentation in AI-Assisted Development. When multiple engineers use different AI agents independently, the work, decisions, and intermediate results are trapped in separate, ephemeral chat sessions or local terminals. This leads to duplicated effort, lost context, and an inability to audit or continue work seamlessly.
2. Target Audience: Software Engineering Teams adopting AI Coding Agents. This includes Engineering Managers seeking visibility into AI-assisted workflows, Senior Developers overseeing complex features built with agent help, and DevOps/Platform Engineers responsible for integrating and securing AI tooling within the development lifecycle.
3. Use Cases:
   • Multi-Agent Code Reviews: An implementation agent writes code, and a separate review agent analyzes it within the same workspace, leaving findings that a human engineer can then adjudicate with a recorded decision.
   • Long-Running Feature Development: A feature goal is set in a workspace. Over days, multiple engineers and their preferred agents contribute incrementally. The workspace maintains full context, allowing any participant to pick up where the last session left off.
   • Auditing and Compliance: For regulated industries, Campfire's immutable history provides an audit trail of which AI agent made which change and which human approved it, addressing accountability concerns in AI-generated code.

Unique Advantages

1. Differentiation: Unlike AI-powered IDEs (which focus on single-user interaction) or complex multi-agent orchestration frameworks (which require defining agent interactions upfront), Campfire is a lightweight, shared workspace. It competes with the lack of structure in current agent use, not with the agents themselves. It's simpler than full orchestration but more structured than disjointed sessions.
2. Key Innovation: The Workspace as a First-Class, Persistent Collaboration Primitive. Campfire's core innovation is treating the shared goal and its evolving context as the central object, decoupled from any specific agent's session. This "harness-independent" approach allows it to work with any agent that can interface with its API, making it a future-proof layer for team-based AI collaboration.

Frequently Asked Questions (FAQ)

1. Does Campfire replace my AI coding agent like GitHub Copilot or Cursor? No. Campfire is a complementary collaboration platform. You continue using Codex, Claude Code, or any other agent inside your editor or terminal. Campfire provides the shared workspace where the outputs and decisions from those agents are gathered and made persistent for the team.
2. How does Campfire handle security and permissions for AI agents? Campfire uses a capability-based security model. When you add an agent to a workspace, you define a specific set of allowed actions (read/write access to certain paths, permission to run specific commands) and explicit denials. This prevents agents from accessing sensitive data or making uncontrolled changes outside their designated scope.
3. Is Campfire an AI agent orchestrator? Campfire does not orchestrate or sequence agent workflows. It does not decide which agent runs next. Instead, it provides a common stateful workspace. Humans and agents decide when to contribute, and Campfire records the results, making the collaboration history visible and persistent.
4. What happens to my Campfire workspace if I stop the server or close my laptop? The workspace and its entire history are persisted independently of any local session. You can close your terminal, and the goal, code changes, findings, and decisions remain. You or another team member can reconnect to the same workspace later to continue the work.
5. Can I use Campfire with my own custom AI agent? Yes. Campfire is open-source and harness-independent. You can integrate any custom AI agent by having it interact with the Campfire CLI or API to pull context from and push results to a workspace, allowing it to participate in the shared collaborative environment.