by subgentic
Coral: The Inbox for Your Coding Agents. You create goals, the Orchestrator creates the tasks, a team of agents picks up the work.
# Add to your Claude Code skills
git clone https://github.com/subgentic/coralGuides for using ai agents skills like coral.
Last scanned: 11/17/2026
{
"issues": [
{
"file": "README.md",
"line": 31,
"type": "remote-install",
"message": "Install command (remote install script piped to a shell — review the source before running): \"curl -fsSL https://github.com/subgentic/coral/releases/latest/download/install.s\"",
"severity": "low"
}
],
"status": "PASSED",
"scannedAt": "2026-11-17T12:49:09.397Z",
"npmAuditRan": true,
"pipAuditRan": true,
"promptInjectionRan": true
}See how coral compares with popular alternatives.
coral is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by subgentic. Coral: The Inbox for Your Coding Agents. You create goals, the Orchestrator creates the tasks, a team of agents picks up the work. It has 132 GitHub stars.
Yes. coral passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/subgentic/coral" and add it to your Claude Code skills directory (see the Installation section above).
coral is primarily written in Go. It is open-source under subgentic on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh coral against similar tools.
No comments yet. Be the first to share your thoughts!
From $49 · 30 days · terms & refunds
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
Automated security scan — not a certification.
# Install on macOS or Linux with one command:
curl -fsSL https://github.com/subgentic/coral/releases/latest/download/install.sh | bash
Running one coding agent for ten minutes in a terminal is easy. Running four, six, or ten agents across a multi-hour or multi-day project is where ad-hoc workflows break down:
Coral turns ad-hoc agent sessions into a structured workflow where your agents report to you:
Needs you) for every permission, question, and deliverable across every running session — instead of babysitting terminal windows.Done when check) and Tasks let you break long-horizon projects into structured steps with review checkpoints — so multi-hour or multi-day work is completed with a high level of control.Instead of checking ten terminal tabs to see who is stuck, you steer everything from one Inbox:
Needs you — one queue for every human decision. Every agent waiting on a tool permission, asking a question (--ask, including Approve/Deny --yes-no prompts), or submitting a plan or file tagged for review appears in one list. Answer in one tap and the agent immediately resumes its turn. The Inbox badge and background tab title show the live count, with optional desktop, Slack, or Discord alerts.Compare and passage comments. Open any plan, report, or file an agent uploaded, select a passage to leave an inline comment (or mark up an image with a pen, arrow, or box), and choose Approve, Approve with a note, or Request changes. When the agent uploads a revision (--new-version-of), Compare highlights the exact lines added and removed between versions.Done when criteria. Give a team a Goal and a single verifiable Done when statement (e.g., "the tests pass and the page loads in under two seconds"). Coral groups every task and artifact under the goal, warns you if review files are still waiting before you close it, and surfaces active goals in Needs you as "looks done" once every task completes.Behind the Inbox is the coordination and memory layer that keeps multi-agent sessions on track over hours or days:
coral-board search and coral-agent history search, while you can search every agent's conversations in Inbox › All chats and Board history.review_pending), and designated reviewers inspect the diff and approve or request changes before dependent tasks unlock.file_read, git_write, shell:<pattern>, or read-only review roles) and Coral translates them into each CLI's native flags and sandboxes.wait, never polling. Workers run coral-board wait for a teammate, task, or commit and end their turn. Coral wakes the agent automatically the moment the reply or dependency lands — zero sleep loops and zero tokens burned waiting.~/.coral/worktrees/<team>) or have agents work in per-task worktrees so your main checkout stays clean. Inspect each agent's uncommitted and committed changes file by file beside its chat.Scan the QR code in t