Canvas sync
AI made building cheap. It didn't make deciding cheap. Mycelium is a Claude Code harness that makes your agent run discovery and weigh evidence before it writes code. It earns the right to start. Built for software, courses, AI tools, and services.
npx -y skills add haabe/mycelium --skill canvas-syncAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Synchronize canvas state across team sessions via git. Ensures all team members see the same product knowledge.
SKILL.md
1.8 KB, as published. Nobody here has run it
Canvas Sync
When to Use
- Starting a new session (pull latest canvas)
- Ending a session (commit canvas changes)
- After team discussion that changes product direction
- When canvas conflicts arise from parallel work
Workflow
Pull Latest
git pull origin main
Then run /mycelium:diamond-assess to see what changed.
Commit Changes
git add .claude/canvas/ .claude/harness/decision-log.md .claude/memory/
git commit -m "canvas: [brief description of what changed and why]"
Conflict Resolution
- Different canvas files: auto-merges (no conflict)
- Same file, different sections: usually auto-merges
- Same field conflicting: the person with more evidence wins — a manual resolution heuristic for the team, not an automated merge. No tooling evaluates evidence weight;
Gated by:a/canvas-mergeprocedure (read both versions, compare provenance depth on overlapping entries, recommend) — not yet built. - When uncertain: discuss as team, log decision
GitOps Principle
The canvas system IS GitOps for product knowledge — git is the single source of truth, changes are declarative (YAML), and updates flow through version control with review. For infrastructure GitOps (ArgoCD, Flux), see JiT tooling detection for Kubernetes/infrastructure-as-code projects.