Pkc Capture Experiment

PKC Capture Experiment

Record what was tried, what was learned, and what it informs.

Process

1. Resolve knowledge root (pkc_common.py resolve-root).
2. Collect: title, hypothesis, method, results, conclusion, related code paths.
3. Write experiments/<slug>.md using template or:

   python3 "${CLAUDE_PLUGIN_ROOT}/scripts/pkc_capture.py" experiment \
     --repo . \
     --title "JWT vs session-cookie spike" \
     --hypothesis "JWT will simplify horizontal scaling" \
     --result "Both workable; JWT slightly simpler ops" \
     --conclusion "Proceed with JWT" \
     --informs user-authentication
   

4. Typed edges:
   • Experiment → Feature/Design/Requirement: rel: informs
   • Decision that followed → Experiment: rel: originates_from (on the DecisionRecord)
5. Refresh experiments/index.md and log.md.
6. Offer to capture a follow-on DecisionRecord via pkc-capture-decision.

Frontmatter essentials

type: Experiment
title: …
hypothesis: …
result: …
conclusion: …
status: completed   # or running | abandoned
verified: true
links:
  - target: /features/….md
    rel: informs

Done when

• Experiment concept written with hypothesis + conclusion
• At least one informs link when a target feature/design is known
• Catalog + log updated