AI Agent Hub
Back to plugins
dsh-agent-replay preview

dsh-agent-replay

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install forrestsweet/dsh-agent-replay

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install forrestsweet/dsh-agent-replay in the DeepSeek Harness terminal; the source repository is https://github.com/forrestsweet/dsh-agent-replay . After restart, a Replay tab appears next to Conversation and Trajectory in any non-empty session.

About this plugin

Harness ships a powerful Trajectory diagnostic view that is ideal for developers debugging event-level details. But when you need to explain what a session actually did to teammates, maintainers, or documentation readers, the raw event ledger is often too verbose and still carries system context and live state that should not leak. Agent Replay fills that communication gap.

It adds a native Replay tab right beside Conversation and Trajectory, turning messages, model replies, tool calls, commands, errors, and timings into a searchable, filterable timeline. You can scrub through events at 1×, 2×, or 4× speed, expand any entry for full detail, copy content, or load earlier history. When the story reads well, one click on Export Shareable Page produces a self-contained, dependency-free HTML file with dark-mode support that any recipient can open in a browser and interact with.

Before export, the plugin automatically drops system and context events as well as hidden assistant reasoning, and redacts macOS/Windows user paths, common GitHub, OpenAI, and Slack-style tokens, Bearer tokens, and key-value credentials. Agent Replay is not a replacement for Trajectory; it is a session-narrative tool designed for anyone who needs to tell a clear, safe story about an agent run and hand it off with confidence.

Screenshots

Use Cases

  • Demonstrate a full agent run to teammates
  • Export a debugged session as a web page for maintainer review
  • Embed an interactive session replay in technical documentation

Best For

  • Agent developers and debuggers
  • Team maintainers and code reviewers
  • Documentation writers and technical readers