Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

fujibee/agmsg — Cross-vendor messaging for CLI AI coding agents — let Claude Code, Codex, Gemini & Copilot talk to e…

Description and use

**Brief Summary (2‑3 sentences)**
fujibee/agmsg is a lightweight, Bash‑driven orchestration layer that lets different CLI‑based AI coding agents (Claude, Codex, Gemini, Copilot, etc.) exchange messages through a shared SQLite store—no background daemon or heavyweight framework required. By turning isolated prompts into a persistent, queryable “agent memory,” it enables repeatable, multi‑agent workflows that can be scripted and version‑controlled like any other shell tool. The project is actively maintained, has strong community traction (≈ 830 ★), and is packaged as a simple CLI/SDK that can be dropped into existing pipelines.

**Value**

  • **Unified communication**: Provides a common “mailbox” for heterogeneous AI agents, removing the friction of custom glue code.
  • **Persisted state**: SQLite‑backed message logs act as a cheap, portable memory layer, making debugging and audit trails trivial.
  • **Zero‑ops deployment**: Because it runs purely as a Bash script with no daemon, it can be added to CI/CD, local dev environments, or edge containers without additional infrastructure.

**Practical Adoption Path**

  1. **Prototype** – Clone the repo, run the provided CLI (agmsg send …, agmsg recv …) against a local SQLite file, and replace existing ad‑hoc curl/openai calls with agmsg messages.
  2. **Integrate** – Wrap each agent’s invocation in a small shell wrapper that pushes its output to agmsg and reads the next input from the queue; commit the wrapper to your repo.
  3. **Scale** – Move the SQLite database to a shared volume (e.g., NFS, Docker volume, or cloud‑attached file) so multiple containers or CI runners can collaborate.
  4. **Extend** – Use the SDK (exposed as simple Bash functions) to embed richer metadata (e.g., task IDs, timestamps) or to plug in a custom persistence layer if needed.

**Production Readiness**

  • **Activity & Community**: Recent commits (as of 2026‑06‑23), 833 stars, 61 forks, and active issue discussion indicate a healthy open‑source project.
  • **Stability**: The core is just Bash + SQLite, both battle‑tested components; there is no long‑running daemon to manage, reducing operational risk.
  • **Integration Simplicity**: Exposes a clear CLI/SDK surface and minimal external dependencies, making security reviews and CI integration straightforward.
  • **Remaining Checks**: Before a full production rollout, verify the license compatibility, run a security scan of the Bash scripts, and confirm an active maintainer is available for critical bug fixes. Once those final reviews are completed, agmsg is a strong candidate for a serious pilot in multi‑agent AI coding pipelines.

Problem

fujibee/agmsg helps turn isolated prompts and tools into repeatable agent workflows.

Outcome

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.

Workflow

  • coordinate multi-agent workflows
  • add tool-use pipelines
  • standardize agent memory

Evidence

Cross-vendor messaging for CLI AI coding agents — let Claude Code, Codex, Gemini & Copilot talk to each other in one team. Bash + SQLite, no daemon, no framework.

Source

  • Record date: 2026-09-10T18:25:58.064Z