transfer

Transfers actionable learnings between sibling projects' .living/learnings.md files.

68|2|Updated Feb 25, 2026
One-click install
npx skills add https://github.com/arjunrajlaboratory/mycelium --skill transfer-arjunrajlaboratory
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: transfer
Source: https://github.com/arjunrajlaboratory/mycelium/tree/main/skills/transfer
Command: npx skills add https://github.com/arjunrajlaboratory/mycelium --skill transfer-arjunrajlaboratory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Insights discovered in one analytical project stay trapped in that project's folder, so sibling projects repeat the same mistakes and rediscover the same workarounds. This Skill scans recent learnings across all subprojects and automatically propagates the ones that generalize. ## Core Features & Use Cases - Cross-project scanning: Finds every .living/learnings.md under a meta-project root and reads only recent entries via tail -80 to stay context-efficient. - Automatic transfer application: Rewrites qualifying learnings for the target project's context and appends them with an auditable source: (auto-transferred by mycelium) tag, skipping near-duplicates. - Audit reporting: Writes a dated report to .living/outputs/knowledge-transfers/ and updates a .last-run timestamp for scheduled runs. - Use Case: A bioinformatics lab with RNA-seq and imaging subprojects runs this as a background subagent so a debugging pattern found in one pipeline is automatically recorded in the others. ## Quick Start Ask the assistant to transfer knowledge and cross-pollinate recent learnings across all sibling projects in the meta-project.

Frequently Asked Questions about transfer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I transfer learnings between multiple projects?

Run the transfer command manually or let it dispatch automatically as a background subagent when more than 24 hours have passed since the last run. It scans each subproject's recent learnings and appends generalizable ones to the relevant target projects.

How does knowledge transfer avoid duplicating existing learnings?

The meta-project's own learnings file serves as the baseline of already-transferred knowledge, and each target file's last 30 lines are checked for near-duplicates before appending. Entries already present in the baseline are skipped.

What kinds of learnings get transferred between projects?

Only actionable items transfer: debugging patterns, shared infrastructure or API insights, workflow improvements, and methodology lessons relevant to related analyses. Project-specific details and trivial connections are ignored.

Does scanning large learnings files slow down the transfer?

No. The workflow never reads full learnings files, using tail -80 on subprojects and tail -100 on the meta-project baseline instead. Total runtime is targeted at under two minutes.

Where is the knowledge transfer audit report stored?

Reports are written to .living/outputs/knowledge-transfers/YYYY-MM-DD.md under the meta-project, with each transfer marked APPLIED. A .last-run timestamp file in the same directory tracks scheduling.