Agent Thread
Share Claude Code and Codex sessions as public links
Last verified:
What is Agent Thread?
Agent Thread is a command‑line tool that lets you export ongoing AI‑coding sessions—such as Claude Code interactions or Codex threads—into public, shareable links with a single command. It captures the context of a live agent session (files, prompts, reasoning traces, and edit plans) and packages them into a hosted view so others can inspect the exact state of the agent’s work without having to recreate the environment. The tool is designed for developers, AI‑assisted coding teams, and technical writers who want to review, debug, or document how an AI agent approached a coding problem or refactor.
Key features include one‑command export that hooks into supported agent environments (Claude Code and Codex‑style threads), automatic serialization of the current session state, generation of a public URL pointing to the exported thread, and minimal friction integrations (Bun, npm, pnpm) so you can run it inside or alongside your editor. It also supports quick installation via common package managers so the same command can be used across different toolchains. The tool is aimed at engineers who want to share fine‑grained, step‑by‑step agent sessions—rather than just final code changes—so teammates can audit reasoning, check for edge cases, or use the exported thread as a living example for onboarding or documentation.
The tool is particularly useful in collaborative development where traceability matters: for PR reviewers to understand how an agent arrived at a solution, for mentors to walk through tutoring sessions with students, or for open‑source maintainers to show how an AI agent actually worked through a complex issue. By turning ephemeral terminal or editor sessions into persistent links, Agent Thread bridges the gap between local experimentation and public reproducibility, making it easier to discuss, reproduce, and iterate on AI‑driven coding workflows.
Agent Thread pricing
Pricing model: Freemium
The tool is distributed as an open‑source CLI package and can be installed and run for free via Bun npm or pnpm; there is no separate paid tier or subscription model shown on the site, and the core functionality (one‑command export to public links) is available at no cost.
Agent Thread pros
- Exports Claude Code sessions to public links in one command
- Exports Codex‑style threads to public links in one command
- Minimal setup with a single CLI command
- Supports multiple package managers (Bun npm pnpm)
- Runs as a global CLI tool without complex configuration
- Generates a single URL representing the full agent session
- Captures the current state of the agent session (files and prompts)
- Simplifies sharing complex AI‑driven coding workflows with others
- Enables teammates to review not just the code but the agent’s reasoning steps
- Facilitates debugging and auditing of AI‑generated edits by sharing links
- Helps maintainers document how an AI agent solved a particular issue
- Supports quick onboarding by sharing example agent sessions as URLs
- Works regardless of underlying agent provider as long as the session format is supported
- Reduces friction when asking for help by linking directly to the agent’s context
- Makes it easy to create a portfolio of AI‑assisted coding sessions as shareable artifacts
Agent Thread cons
- Limited to specific agent environments (Claude Code Codex‑style formats)
- Requires the agent session to be running when you run the export command
- Tied to the current state of the session so older states may not be captured
- Public links may expose sensitive code or configs if not handled carefully
- No built‑in access controls or permissions on the exported threads
- No version history or diff views directly inside the exported link
- Relies on the underlying agent’s internal state format which can change
- No offline or self‑hosted viewing mode for the exported sessions
- Does not automatically notify collaborators when a new thread link is shared
- Lacks annotation or commenting tools on the exported thread view
Frequently asked questions about Agent Thread
What is Agent Thread
Agent Thread is a command‑line utility that lets you export live AI‑coding sessions—such as Claude Code sessions or Codex threads—into public, viewable links with a single command. It captures the current state of the agent session (files, prompts, and reasoning traces) and surfaces them at a URL so others can inspect how the agent approached a coding task without having to recreate the local environment.
Which agent environments does Agent Thread support
Agent Thread currently supports exporting sessions from Claude Code and Codex‑style agent threads. It is designed around these specific environments and may not work out of the box with arbitrary AI agents unless their session format is compatible with the tool’s internal export logic.
How do I install Agent Thread
Agent Thread can be installed using any of several common package managers: run bunx agent-thread@latest, npx agent-thread@latest, or pnpx agent-thread@latest to execute it globally without installing it permanently, or add it as a local dev dependency if you prefer to pin the version in your project.
What do I get when I run Agent Thread
When you run Agent Thread in a supported agent session, it serializes the current state of that session (including active files, prompts, and the agent’s reasoning) and generates a public URL pointing to a hosted view of that exact thread so you can paste it into chats, issues, or documentation for others to review.
Is there a cost to use Agent Thread
There is no monetary cost to use Agent Thread; it is provided as an open‑source CLI package that can be installed and run via package managers like Bun npm or pnpm, and the core export functionality is available for free without any paid upgrade or subscription.
Can I restrict who sees my exported thread links
The website does not describe any built‑in access controls or authentication for the exported links; threads are made publicly viewable at the generated URL, so you should treat them as public and avoid sharing threads that contain sensitive code, API keys, or private configuration.
How permanent is an exported thread link
An exported thread link is treated as a persistent snapshot of the session state at the moment you ran the command; the page remains accessible as long as the hosting backend continues to serve it, but there is no documented versioning or long‑term retention guarantee on the service itself.
Can I export multiple points in the same session
You can run Agent Thread multiple times during the same session to capture different states, producing separate links for each export; each link will reflect the session state at that specific moment rather than continuously updating as the session evolves.
What information is included in the exported thread
The exported thread typically includes the files that were active in the session, the prompts and instructions sent to the agent, and the agent’s reasoning or planning steps as it generated or modified code, allowing others to follow the full context of how the agent arrived at its changes.
Is there an API or advanced configuration for Agent Thread
The website does not advertise a public API or complex configuration options; the tool is presented as a simple CLI with a one‑command export pattern, aimed at ease of use rather than low‑level extensibility or programmatic integration beyond the basic export command.