iloom
Scale understanding, not just output — AI code you and your team can trust
Note: Supports macOS and Linux (including WSL). Native Windows is not yet supported.
A VS Code extension that turns AI chaos into structured workflows. Every decision documented. Every task isolated. Every PR reviewable.
The Problem
Nobody wants to review your AI's work. Including you.
AI coding assistants are powerful, but working with them introduces new challenges:
- Invisible reasoning: You can't review what you can't see—AI makes decisions without documenting why
- Team trust: You don't trust your AI's code. Neither does your team.
- Context decay: AI starts hallucinating or forgetting requirements mid-task
- Branch chaos: Parallel tasks create port conflicts, stale branches, and environment collisions
- Mental overhead: You're juggling 5 branches, but it's your brain that has the merge conflict
iloom solves these through structured workflows, visible & persisted reasoning, and complete isolation for each task.
How iloom Works
Starting a loom doesn't just create a branch. It launches a multi-agent workflow that:
- Aligns on requirements before any code is written
- Surfaces assumptions and creates a structured plan
- Captures key decisions, risks, and insights along the way
- Persists everything in your issue tracker—analyses, plans, results and a summary when you're done
- Executes with full context in an isolated environment
- Verifies tests and lint before completion
The result: AI-generated code with a clear paper trail, isolated from your other work, ready for meaningful review.
Requirements
The iloom CLI is required. This extension is a UI layer for the iloom CLI tool.
Install the CLI before using this extension:
npm install -g @iloom/cli
Prerequisites:
- macOS or Linux (including WSL)
- Git
- Node.js 18+
- GitHub CLI (
gh) — for GitHub integration
- Linear CLI (optional) — for Linear integration
- Claude Max subscription (strongly recommended)
Quick Start
- Install the extension from the VS Code Marketplace
- Install the CLI:
npm install -g @iloom/cli
- Open a project folder in VS Code
- Find iloom in the left-hand activity bar—look for the lightbulb
- Set up your project: You'll be prompted to configure iloom for this project—main branch, issue tracker, and more
- Create your first loom: Click
+ New Loom and enter an issue number
Core Workflows
Create a Loom
Browse open issues from GitHub or Linear directly in the extension, or enter a PR number or free-form description. The new loom modal lets you search and filter issues by ID or title, so you don't need to leave VS Code to find the right task. iloom then takes you through a structured workflow:
- Requirements alignment: AI analyzes the issue and surfaces questions before coding
- Research & analysis: Deep dive into the codebase to understand context and constraints
- Plan creation: A detailed implementation plan is posted to your issue tracker
- Isolated execution: Work happens in a separate worktree with its own resources
- Verification: Tests and lint run before completion
Supports browsing open issues, issue numbers (#123, ENG-123), PR numbers, or branch names.
Manage Looms
The iloom explorer panel in the activity bar shows all active workspaces across your projects:
- Type indicators: Issue, PR, or branch-based looms
- Session status: See when your AI needs your attention
- Quick actions: Switch, finish, rebase, or remove looms
Complete a Loom
When your work is done, click the 🏁 finish flag to wrap up:
- Validates tests and lint pass
- Rebases onto main (or parent loom), resolving conflicts with AI assistance
- Merges or creates PR based on your configured merge behavior:
local (default): Merge directly into main branch locallygithub-pr: Push and create a pull requestgithub-draft-pr (default when contributing to open source): Create draft PR on start, mark ready on finish
- Cleans up the worktree and closes the workspace window
Epics & Swarm Mode
For large features that span multiple issues, iloom provides a full epic workflow — from planning through parallel execution.
Plan an Epic
Break a large feature into structured child issues with dependency relationships:
- Click Plan Epic in the iloom explorer, or run
iloom: Plan Epic from the Command Palette
- Enter an epic description or issue number
- AI analyzes the scope and creates a structured breakdown — each child issue gets its own isolated loom
- Dependency relationships between child issues are captured automatically
Swarm Mode
After planning, iloom can execute the entire epic autonomously:
- Parallel AI agents — one per child issue, each working in its own isolated worktree
- Respects the dependency DAG — blocked issues wait for their prerequisites to complete
- Automatic merging — completed work is rebased and merged back into the epic branch
- Graceful failure handling — if one child fails, unrelated work continues
- Live progress tracking — watch it all unfold in the Epic Board
Epic Board
The epic board gives you a live view of progress across all child issues. Switch between two views:
- Board view: A Kanban board showing issues organized by status — Pending, In Progress, Code Review, and Done
- Dependencies view: A DAG (directed acyclic graph) showing how child issues depend on each other, with status-colored nodes and dependency edges
Isolation Features
Each loom runs in complete isolation from your other work:
| Resource |
How iloom Isolates It |
| Git worktree |
Separate filesystem—no stashing, no branch switching, no conflicts with other looms |
| Database branch |
Schema changes isolated from main and other looms (Neon integration) |
| Unique port |
Web apps run on deterministic ports (base port + issue number)—no more "port taken" mysteries |
| Isolated binary |
CLI tools get workspace-specific binaries (e.g., my-cli-25 when working on issue #25) |
This means you can have five looms running simultaneously, each with their own dev server, database state—without interference.
Child Looms
Create nested looms within a loom — useful for fixing bugs discovered mid-task or manually splitting work into smaller PRs. Child looms inherit the parent's database state and branch.
For automated orchestration of child issues, see Epics & Swarm Mode above.
Open Source Contributions
Maintainers shouldn't be overwhelmed by unexplained AI-generated code. With iloom, your analysis and plans post directly to the PR—giving reviewers full context on why the code looks the way it does.
il contribute n8n-io/n8n
This forks the repo, clones your fork, configures iloom, and sets up draft PR mode automatically.
Run il contribute with no arguments to contribute to iloom itself.
Features
iloom Explorer
The activity bar panel provides a central view of all your active looms:
| Feature |
Description |
| Project grouping |
Looms organized by project |
| Type icons |
Visual distinction for issues, PRs, and branches |
| Session status |
Know which loom needs your attention and why |
| Quick switch |
Click to open any loom workspace |
Commands
Access via Command Palette (Cmd+Shift+P / Ctrl+Shift+P):
| Command |
Description |
iloom: Start New Loom |
Create a new loom from issue or description |
iloom: Quick Switch Loom |
Fast switch between active looms |
iloom: Refresh |
Refresh the iloom explorer |
iloom: Settings |
Configure iloom for the current project |
One Shot Mode
When creating a loom, you can select a "One Shot Mode" to adjust how the AI operates:
| Mode |
Description |
| Default |
Standard workflow with all checkpoints |
| Skip Reviews |
Bypass review phases for faster iteration |
| YOLO Mode |
Full autonomy—AI proceeds without permission checks. Use with caution. |
Recap Panel
Track your workflow progress with the Recap panel:
- Timeline of insights, decisions, and artifacts
- Links to issue comments, new issues and PRs created during sessions
- Risk and assumption tracking
AI Integration
iloom integrates with AI coding assistants to provide workflow context:
- Session status indicators: See when Claude Code or other AI tools are actively working
- Context sharing: Looms provide structured context about the task at hand
- Artifact tracking: Automatically tracks issue comments, PRs, and other outputs
- Conflict resolution: AI helps resolve merge and rebase conflicts
- Commit messages: AI generates helpful, context-aware commit messages
Privacy & Telemetry
iloom collects anonymous usage analytics to improve the product. Here's exactly what that means.
What We Collect
- Usage events: Which panels you open and which views you use (e.g., "opened epic board", "switched loom")
- Environment info: Extension version, OS, and VS Code version
- No personally identifiable information is ever collected
What We Don't Collect
- GitHub, Linear, or Jira usernames or emails
- Repository names or URLs
- Issue titles, descriptions, or content
- File paths or code content
- Branch names
- AI analysis or plan content
- Anything that could identify a specific project or person
How to Opt Out
iloom has two products — the VS Code extension and the CLI (il) — each with independent telemetry controls. Disabling one does not affect the other.
| Product |
How to Disable |
Scope |
| VS Code extension |
Set iloom.telemetry.enabled to false in VS Code settings, or set VS Code's built-in telemetry.telemetryLevel to "off" |
Disables extension telemetry only |
CLI (il) |
See il config or CLI documentation |
Disables CLI telemetry only |
| Both |
Opt out in both places above |
Disables all iloom telemetry |
When extension telemetry is disabled, the analytics SDK is never initialized — no data is sent at all.
Where Data Goes
Anonymous data is sent to PostHog, a privacy-focused analytics platform.
Shared Anonymous ID
The extension and CLI share an anonymous ID (a random UUID) stored at ~/.config/iloom-ai/telemetry.json. This lets us understand the full product journey without knowing who you are. Opting out of telemetry in either product stops that product from sending events, but does not delete the ID file.
FAQ
Q: Why do I need the CLI installed separately?
The extension is a thin UI layer. All git operations, worktree management, and issue tracker integration are handled by the CLI. This separation allows CLI updates without extension reinstallation.
Q: What issue trackers does iloom support?
iloom supports GitHub Issues and Linear. JIRA support is on the roadmap.
Q: What AI tools does iloom support?
iloom is focused on supporting the Claude Code command line tool.
Q: How do looms differ from regular branches?
Looms are git worktrees—separate working directories that share the same repository. Combined with isolated ports, database branches, and CLI binaries, you can work on multiple issues simultaneously without any resource conflicts.
Support
For extension issues, feature requests, and questions:
github.com/iloom-ai/iloom-vscode-support/issues
For CLI issues, visit github.com/iloom-ai/iloom-cli/issues.
License
© 2026 iloom AI. All rights reserved.
This extension is proprietary software. The iloom CLI is separately licensed under BSL-1.1.
This software is provided "AS IS", without warranty of any kind.
