Feedback loops for agentic development, debugging, and user support.
AgentLoops unifies bugs, feature requests, and user feedback into one
agent-friendly resolution loop — with a CLI, a dashboard, and an MCP
server so AI coding agents can triage, track, and resolve issues
alongside your team.
📦 npm: @stevenvincentone/intidev-agentloops🧩 TypeScript & JavaScript🤖 MCP server included⚖ MIT license
One ticket surface for bugs, features, and feedback
Most teams run separate systems for bug tracking, feature requests, and
user support — and AI coding agents end up stitching them together by
hand. AgentLoops takes a different approach: it gives
every kind of feedback — a bug report, a feature idea, a support ticket,
an incident — a single, structured lifecycle that both people and agents
can read and act on directly.
Tickets move through clear status transitions —
triaged → active → resolved / deferred / reopened — and each one
gets a machine-readable alias like ISSUE-000001. State is
stored in a simple .agentloops/state.json file by default,
with pluggable backends including filesystem, in-memory, and PostgreSQL
for larger teams.
"An open-source toolkit for tracking issues, features, and user feedback through an agent-friendly resolution loop."
⚙
Key features
Built for human + agent collaboration
🔁
Unified ticket lifecycle
Bugs, features, support requests, and incidents all flow through one consistent, agent-readable lifecycle.
🧠
Prior-art detection
Automatically surfaces related historical tickets when a new one is created — so agents don't duplicate work.
🕸️
Pattern & convergence detection
Clusters recurring issues into shared root-cause Patterns, and flags when independent reports converge on the same cause.
🤝
MCP server for AI agents
A Model Context Protocol server gives coding agents read/write access — they can triage, resolve, and write up tickets natively.
💻
Full-featured CLI
~30 commands for creating, resolving, reopening, deferring, and grouping tickets — scriptable for any workflow.
📊
Dashboard UI
Zero-dependency static HTML dashboard, or run it live — plus optional React components for embedding in your own app.
🔗
GitHub Issues sync
Optionally mirror tickets to GitHub Issues for public visibility and contributor workflows.
🛡️
Regression guards
Track resolved issues over time and get flagged if a fix appears to regress — closing the loop for real.
📦
Getting started
Install it in minutes
AgentLoops installs globally via npm and initializes itself inside any
project. From there, your team — and your AI agents — can start creating
and resolving tickets right from the command line.
Install & initialize
# install the CLI globally
npm install -g @stevenvincentone/intidev-agentloops
# set up AgentLoops in your project
agentloop init
Create and manage tickets
# report a bug
agentloop create --title "Login fails on Safari" --type bug
# list active tickets
agentloop list --status active
# resolve a ticket with a note
agentloop resolve ISSUE-000001 --note "Fixed in PR #42"
Prefer a UI? Generate the static dashboard with agentloop dashboard,
or run it live to browse tickets, patterns, and resolution history in
the browser. Teams that want deeper integration can install
@stevenvincentone/intidev-agentloops-react to embed the
dashboard components directly into an existing internal tool.
For AI coding agents, the bundled MCP server exposes the same
functionality over the Model Context Protocol — so an agent can search
for prior art, open a ticket, cluster it into a Pattern, and write up a
resolution, all without leaving its own loop.
Install IntiDev AgentLoops, run agentloop init, and start
turning scattered bug reports and feature ideas into one resolution
pipeline your whole team — human and AI — can work from.