paradigms / playbook.yaml
DranFren's picture
Upload 5 files
1b21338 verified
playbook:
metadata:
name: "ACE–CPT Playbook"
version: "0.1"
items:
- id: "PB-001"
type: "checklist"
title: "Stillness preflight"
content: |
Restate task (1 line). List 2–3 success criteria. One salient risk. Name initial paradigm lens (data/method/normative).
tags: ["meta","stillness"]
helpful: 0
harmful: 0
- id: "PB-002"
type: "procedure"
title: "Dual-source / unit-test verify"
content: |
For claims affecting safety/finance/medical/legal: confirm with an independent source or unit test before final Answer.
tags: ["verification"]
helpful: 0
harmful: 0
- id: "PB-003"
type: "checklist"
title: "Anthropomorphization & spec-gaming sentinel"
content: |
(a) Replace mind-language with mechanism; (b) check rubric-gaming; (c) add uncertainty note if incentives misalign.
tags: ["safety","bias"]
helpful: 0
harmful: 0
- id: "PB-004"
type: "heuristic"
title: "Paradigmatic surfacing (2.11–2.17)"
content: |
Ask: ontology, evidence, method, objective, normative values, institutional forces, closure/blind spots.
tags: ["cpt"]
helpful: 0
harmful: 0
- id: "PB-005"
type: "failure_mode"
title: "Context collapse via over-summarization"
content: |
Avoid replacing many atomic items with a single generic prompt; use deltas and dedupe instead.
tags: ["hygiene"]
helpful: 0
harmful: 0