skills.rest
ZanzyTHEbar
ZanzyTHEbarCOMMUNITY
Explore all skills by this author →
𝕏@ZacariahHeimGitHub

workflow-from-chats

Extract durable working preferences from OpenCode or Cursor chat transcripts into skills and rules.

3|Updated Apr 8, 2025
Search Tags:#personalization#opencode#cursor#agent rules#chat transcripts#workflow mining#preference extraction
One-click install
npx skills add https://github.com/ZanzyTHEbar/dragonarchy --skill workflow-from-chats-zanzythebar
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: workflow-from-chats
Source: https://github.com/ZanzyTHEbar/dragonarchy/tree/main/packages/opencode/.config/opencode/skills/workflow-from-chats
Command: npx skills add https://github.com/ZanzyTHEbar/dragonarchy --skill workflow-from-chats-zanzythebar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Valuable working preferences and corrections get buried in past AI chat sessions and are never reused. This Skill mines recent OpenCode or Cursor transcripts to convert repeated feedback into persistent skills, rules, or workflow documentation. ## Core Features & Use Cases - Preference Extraction: Scans transcripts for explicit preference markers like "I prefer", "always", "never", and workflow corrections, then rates each finding by confidence (strong, medium, weak, contradicted). - Artifact Generation: Clusters preferences by workflow shape (shipping, review, debugging, delegation) and produces the right artifact: a new skill, a rule, a workflow doc, or a Memory Bank update. - Privacy-Safe Synthesis: Cites only parent conversations, never exposes local transcript paths, secrets, or private chat content. - Use Case: After two weeks of correcting your AI assistant on PR review style, ask it to mine recent chats and generate a reusable review skill encoding your preferences. ## Quick Start Ask the assistant to analyze your recent OpenCode or Cursor chats from the last 7 days and turn your recurring preferences into a reusable skill or rule.

Frequently Asked Questions about workflow-from-chats

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

FAQPage Schema
How do I turn my AI chat history into reusable workflow rules?▼

Run this Skill against your recent OpenCode or Cursor transcripts. It scans for explicit preferences and corrections, clusters them by workflow type, and drafts a skill, rule, or workflow doc capturing the reusable guidance.

How to extract user preferences from OpenCode or Cursor chats?▼

The Skill builds a transcript inventory, scans for markers like "I prefer", "always", and "never", then extracts preference atoms with trigger, decision rule, and confidence rating. It defaults to the last 7 days of chats.

Does chat transcript analysis expose my private conversations?▼

No. The Skill cites only parent conversations and never exposes local transcript paths, secrets, customer data, or private chat content. Subagent transcripts are used as evidence but never cited directly.

What happens if transcript access is unavailable?▼

The Skill falls back to the current conversation and any excerpts you explicitly provide. It will not invent a broader evidence corpus or fabricate preferences from unavailable history.

When should mined preferences not become a skill or rule?▼

Situational, stale, or low-confidence observations produce no artifact. Contradicted evidence, where transcripts point in incompatible directions, requires asking the user before writing any files.

Related Skills

Explore More Software Engineering Skills→
kage1020kage1020

prompt-review

Analyze past AI agent dialogue histories and write a dated report.

Community
Advanced
cursorcursor
6.6k

automate-me

Drafts a personal mode skill capturing user working conventions from transcript mining.

Official
Advanced
VectorInstituteVectorInstitute

fix-merge-conflicts

Resolve git merge conflicts across dependency, source, and configuration files.

Official
Advanced
TyrizxTyrizx

create-instructions

Create .instructions.md files that codify project rules and coding conventions.

Official
Intermediate