amplifthq/opentag

What it solves

OpenTag allows teams to integrate local coding agents into their existing collaboration platforms (like Slack, GitHub, or Discord) without moving the conversation to a separate AI tool. It solves the problem of fragmented context by keeping the agent's work loop within the original work thread, while ensuring that agent actions are governed, auditable, and executed locally for privacy and security.

How it works

OpenTag acts as a local dispatcher and runner that connects a collaboration platform to a coding agent. When a user mentions the agent in a thread, OpenTag curates a context packet and triggers a local coding agent (via the Agent Client Protocol - ACP). The agent performs the work on the local machine and returns concise artifacts or proposed actions to the thread. Users can then approve or reject these actions through "action receipts" directly in the chat interface.

Who it’s for

It is designed for developers and engineering teams who want to leverage AI coding agents (such as Claude Code, Cursor, or Codex) while maintaining strict control over their code and execution environment through a local-first approach.

Highlights

• Multi-platform support: Integrates with Slack, GitHub, GitLab, Linear, Lark, Telegram, Discord, and Microsoft Teams.
• Local-first execution: Coding agents run on the user's machine, ensuring credentials and code stay private.
• Governed work loops: Uses action receipts to require human approval for system-of-record mutations (e.g., creating a pull request).
• Audit trails: Maintains a local agent work ledger recording every event, decision, and artifact produced during a run.
• Executor-neutral: Supports various ACP-compatible agents including Claude Code, Cursor, and OpenCode.