Transfer anonymized prevention lessons from another AI Factory project into current skill-context.
npx skills add https://github.com/lee-to/ai-factory --skill aif-transfer
Please help me install this Agent Skill. Skill: aif-transfer Source: https://github.com/lee-to/ai-factory/tree/main/skills/aif-transfer Command: npx skills add https://github.com/lee-to/ai-factory --skill aif-transfer
When starting a new project similar to a previous one, hard-won lessons captured in the old project's fix patches are lost. This Skill reuses those prevention lessons without leaking the source project's identity, paths, or proprietary identifiers into your current project.
.ai-factory/patches directory, extracts prevention points, and anonymizes all source identifiers before use./aif-evolve workflow, which applies approved improvements to skill-context and writes an evolution log.Run /aif-transfer with the path to a reference AI Factory project, for example /aif-transfer "/path/to/reference-project" all, then approve the proposed improvements.
High-intent search queries and answers about installing and using this skill.
Run /aif-transfer with the source project path, for example /aif-transfer "/path/to/reference-project" all. The skill reads the source project's fix patches, filters relevant prevention lessons, anonymizes them, and runs the /aif-evolve workflow for your approval.
No. Source patches are never copied into the current project's patches directory. Only anonymized prevention lessons that pass relevance and privacy checks are proposed through /aif-evolve, and nothing is written without your explicit approval.
No. The source project is strictly read-only for the entire run. The skill only reads the source config, description, architecture, and patch markdown files, and never calls Write or Edit on any path under the source root.
If no prevention candidate survives the relevance gate, the skill reports only the analyzed and rejected counts and stops. No evolution log is created and no skill-context files are changed.
It builds an in-memory denylist of source names, paths, URLs, and identifiers, rewrites candidates with neutral roles or verified current-project equivalents, and scans all outputs before and after writes. Source identity never appears in artifacts or reports.
The current project must be initialized with /aif and have a description artifact, the aif-evolve skill must be available, and the source project must contain a .ai-factory directory with at least one patch file.
ichinyaBootstrap AI Factory projects by creating config.yaml and rules/base.md.
ckallumIdentify and implement deterministic prevention for Claude Code workflow mistakes and anti-patterns.
ArtemYurovAnalyze patches and codebase patterns to iteratively refine AI Factory skills.
YaroslavKomarovAnalyzes project patches and context to generate SKILL.md improvement plans.