Case Proof Check: What Can You Claim

Paste project notes, supporting excerpts and publication permission. Decide whether an outcome case, a process story or a hold-for-evidence path is justified before writing polished copy.

What you receive

• A clear outcome/process/hold decision and main credibility gap.
• A claim-by-claim table with exact excerpts, support level, permission and limitations.
• A complete supported case/process post, estimated short script and truthful CTA when possible.
• A sentence-to-evidence map and up to five checks.

What to paste

What actually changed, any result you want to claim, the supporting excerpts or metrics with source/period, and what is explicitly allowed for publication. A process story does not require a success metric. Unknown permission stays unknown. Paste ordinary notes; no JSON preparation. English and Traditional Chinese quick starts and worked examples are included. Ask for your preferred output language.

Synthetic example

A permitted project note says: We combined scattered inquiry fields into one checklist. No revenue data is supplied.
Decision: process story only. The note supports reorganizing the form, not doubled revenue or causality. The helper only checks reference/permission structure: every claim remains pending semantic review and not publication-ready. A separate model support table explains what can be drafted. Unsupported outcomes are omitted.

Download and setup

A one-time standard Skill ZIP, not hosted Capafy chat. Install in your own compatible Agent Client; your AI account/model and fees are separate. The optional helper needs Python 3.10+ and only the standard library. Main results are Markdown/plain text; JSON is optional advanced input and evidence.

Data and limits

Scripts only read/write local files: no network, API calls, telemetry, messages or automatic publishing. Your selected AI client/provider processes pasted text under its own terms and fees. Redact private data and review drafts. Semantic support classification is model interpretation, not independent fact checking or legal permission clearance. No invented testimonials, statistics or promises of sales, revenue or error-free model behavior.

Direct quotation checks

Checks proposed direct quotations against supplied excerpts exactly. A match still requires context and permission review; unsupported claims and altered testimonials stay out of public drafts.