hypothesis-tracking
Track and manage research hypotheses with confidence levels. Use when the user asks about prior research, wants to validate claims, or queries what is known about a company.
SKILL.mdskills/hypothesis-tracking/SKILL.md
GitHubHypothesis Tracking
Purpose
Use this skill to manage durable research memory for a company. Hypotheses are a working memory layer, not final truth.
Workflow
Retrieve hypotheses before starting duplicate research.
Present confidence clearly: untested, validated, invalidated, or confirmed.
Update confidence only when new evidence exists.
Keep evidence strings short, factual, and source-oriented.
Operating Rules
Distinguish known facts from open hypotheses.
Preserve the evidence trail when confidence changes.
Prefer updating an existing hypothesis over creating duplicates.
Expired or weak hypotheses should be re-validated before being presented as current.
Example
get_hypotheses(company="ExampleCo", topic="technology")
save_hypothesis(company="ExampleCo", hypothesis_id="h_001", confidence="validated", evidence="CTO interview confirms AWS usage")