evidence-classification

id	skill-evidence-classification
type	agent-skill
status	reviewed
created	2026-05-30T00:00:00.000Z
updated	2026-05-30T00:00:00.000Z
name	evidence-classification
description	Classify claims as fact, opinion, marketing, hypothesis or open question before using them in agent memory or standards.
version	0.1.0
topics	["agent-skills","evidence","assessment"]
tools	["Pritha","Markdown"]
sources	["04_standards/expert-information-assessment.md","07_workflows/expert-information-assessment.md","08_templates/assessment.md"]
related	{"workflows":["07_workflows/agent-skill-pack-selection.md"]}
source	pritha-memory
source_paths	["04_standards/expert-information-assessment.md","07_workflows/expert-information-assessment.md","08_templates/assessment.md"]
review_status	reviewed
trust_level	local-reviewed
requires_toolsets	["filesystem","markdown"]
risk_level	low
tags	["evidence","assessment","standards"]

Evidence Classification

When to Use

Use when converting a source, user note, transcript, article, repository or external claim into curated agent knowledge.

Procedure

1. Separate facts, opinions, marketing claims, hypotheses and unknowns.
2. Prefer primary sources for volatile technical claims.
3. Record source date, retrieved date, verified date and version context when available.
4. Compare the claim with existing briefs, reviews, decisions and standards.
5. Promote only well-supported, reusable conclusions into standards or decisions.

Pitfalls

• Do not treat generated wiki pages as evidence.
• Do not collapse "interesting" and "adopt now" into the same recommendation.
• Do not hide weak evidence behind confident language.

Verification

• Evidence quality is stated.
• Alternatives or trade-offs are named.
• Open questions remain explicit.
• Techscope/Agents Mother fit is recorded when relevant.