Record session learnings and outcomes to docs/switch knowledge stores.
npx skills add https://github.com/Untangling-Systems/switch --skill sw-compound
Please help me install this Agent Skill. Skill: sw:compound Source: https://github.com/Untangling-Systems/switch/tree/main/skills/sw-compound Command: npx skills add https://github.com/Untangling-Systems/switch --skill sw-compound
Teams and change leads lose institutional memory after diagnoses and interventions because learnings, outcomes, and cross-initiative patterns are not captured in a discoverable, structured way. This Skill provides a lightweight, guided workflow to extract non-obvious insights from diagnosis and intervention sessions, detect conflicts with prior decisions, and persist actionable learnings into the repository so future sessions start smarter.
Use the sw:compound skill to record a specific learning or outcome from your recent diagnosis, intervention, or implementation session so it is persisted to the project's switch knowledge stores.
High-intent search queries and answers about installing and using this skill.
Capture session learnings by extracting non-obvious insights and persisting them into project knowledge stores under docs/switch. This Skill appends records to current-diagnosis.md, updates intervention notes, or creates pattern files to preserve institutional memory.
Record Rider/Elephant/Path diagnosis insights by using a guided workflow that prompts you for specific, actionable outcomes. It enforces reference quality and saves the intervention learnings directly to your repository's knowledge stores.
Conflict detection works by searching existing learnings in docs/switch when saving new intervention outcomes. If new insights conflict with prior decisions, it offers explicit options to update, archive, keep both, or discard the conflicting entries.
Yes, you can automate saving cross-initiative patterns using the optional pipeline mode. This mode skips interactive AskUserQuestion prompts, automatically flags conflicts for later review, and tags saves for subsequent inspection.
Yes, context detection reads recent files in docs/switch to determine if the session is post-diagnosis, post-intervention, outcome recording, or pattern discovery, requiring existing files to accurately guide the learning extraction workflow.
Untangling-SystemsGenerate prioritized Rider, Elephant, and Path intervention plans from a diagnosis file.
Untangling-SystemsAnalyze change challenges with the Rider/Elephant/Path framework and write a markdown diagnosis file.
Untangling-SystemsGenerate Elephant intervention plans and save them to docs/switch/interventions.
severeonCapture session learnings and update PROGRESS.md from git history.
