fact-checker

1# Fact Checker
2 
3Verify factual claims in documents and propose corrections backed by authoritative sources.
4 
5## When to use
6 
7Trigger when users request:
8- "Fact-check this document"
9- "Verify these AI model specifications"
10- "Check if this information is still accurate"
11- "Update outdated data in this file"
12- "Validate the claims in this section"
13 
14## Workflow
15 
16Copy this checklist to track progress:
17 
18```
19Fact-checking Progress:
20- [ ] Step 1: Identify factual claims
21- [ ] Step 2: Search authoritative sources
22- [ ] Step 3: Compare claims against sources
23- [ ] Step 4: Generate correction report
24- [ ] Step 5: Apply corrections with user approval
25```
26 
27### Step 1: Identify factual claims
28 
29Scan the document for verifiable statements:
30 
31**Target claim types:**
32- Technical specifications (context windows, pricing, features)
33- Version numbers and release dates
34- Statistical data and metrics
35- API capabilities and limitations
36- Benchmark scores and performance data
37 
38**Skip subjective content:**
39- Opinions and recommendations
40- Explanatory prose
41- Tutorial instructions
42- Architectural discussions
43 
44### Step 2: Search authoritative sources
45 
46For each claim, search official sources:
47 
48**AI models:**
49- Official announcement pages (anthropic.com/news, openai.com/index, blog.google)
50- API documentation (platform.claude.com/docs, platform.openai.com/docs)
51- Developer guides and release notes
52 
53**Technical libraries:**
54- Official documentation sites
55- GitHub repositories (releases, README)
56…