universelab SCI · CLI · BIO

EXP-001 · active

Harness Canopy

In a forest, the canopy is where the crowns touch — separate trees, one living layer.

A Rust daemon and terminal UI that runs alongside your AI agents. It gives them persistent memory across sessions, background scheduling on cron and file events, a knowledge graph that learns from every run, and a sync protocol so multiple agents stop colliding in the same workspace.

Install
curl -fsSL https://get.univerlab.org/canopy | sh
Supported platforms
  • antigravity
  • blackbox
  • claude
  • cline
  • codex
  • continue
  • copilot
  • cursor
  • gemini
  • kilo
  • kiro
  • mimocode
  • mistral
  • opencode
  • qwen
Genesis · TERRA 2026 · Sol 79

The one that started it all.

It started as a folder of skills for work. Then I noticed what nobody was talking about: agent harnesses shipped a headless mode, just sitting there unused. I wired cron jobs to fire tasks through it — too much for a skill, and the models of the day choked on the instructions. So it became an MCP: task-trigger. It worked, but it ran blind in the background; only the agent ever saw what happened. Not enough. I killed it and built a TUI — then scheduling, memory, sync, identities, and a new name. Canopy. By then the twist was complete: Canopy was building Canopy, and everything else in this lab.

What it does

Memory that persists

Every session writes facts and patterns to a project-scoped knowledge graph. The next session reads them. Agents stop re-explaining the same codebase to themselves.

Background scheduling

Agents run on cron schedules or file-change triggers — not just on demand. A daemon watches the workspace so you don't have to babysit.

Multi-agent sync

Agents declare their mission, report stability, and broadcast messages. The workspace "vibe" is visible before anyone touches a file. No more silent collisions.

Loop engine

Watch a loop assemble itself

  1. 01

    A spec meets an agent

    Work enters as a spec — role, what, how. An implementer node picks it up on whichever harness you choose.

  2. 02

    Deterministic gates

    A check node runs your real commands — build, lint, tests. Red routes back to the implementer; only green moves forward.

  3. 03

    Review on a second brain

    A different harness reviews the diff against the spec and commits. Same-vendor blind spots stay out of your branch.

  4. 04

    Failure is routed, not lost

    If the implementer dies mid-run, a resilience node triages it: glitches retry now; a quota death schedules the loop to wake itself at the exact reset time.

  5. 05

    Ensemble mode

    Fan the spec out to several models in parallel, wait for every proposal, and let an arbiter distill the consensus before a single line is implemented.

Every node runs on the harness you pick. Mix vendors freely — the graph is yours to edit.

Loop engine

DAG-based automation

Define workflows as directed acyclic graphs: specs flow through agent, check, and gate nodes with pass/fail routing. Automate bug fixing, code review, and multi-step tasks.

Background execution

Loops run autonomously in the background — implement, verify, review, commit. Each node has timeouts, retries, and a resilience agent that diagnoses failures.

Human-in-the-loop

When automation hits a wall, the resilience agent reports a blocker and pauses. You decide; the loop resumes when you're ready.

FAQ

FAQ

What does Canopy actually do?
Canopy orchestrates work across different AI coding harnesses — Claude, Codex, or any agent that runs in a terminal. It lets you use all your free tiers without learning each platform's commands, config, or MCP parsing. One daemon, all agents.
How do I share work between Claude and Codex?
Canopy gives each agent a shared knowledge graph and sync protocol. When Claude finishes a task, the facts and patterns it discovered are available to Codex in the next session. No manual context copying.
Can I run AI agents on a schedule?
Yes. Canopy fires agents on cron schedules or file-change triggers via a background daemon. Set the schedule once; the daemon watches the workspace and runs tasks automatically.
How is Canopy different from just using Claude Code or Codex directly?
Those are individual harnesses. Canopy is the layer that connects them — shared memory, background scheduling, and multi-agent coordination. You keep your agents; Canopy adds the infrastructure.