Overclaim Detect

Via MCP

Use the rigorous.overclaim tool with parameters:

• file: path to manuscript
• sections: optional comma-separated section filter

What It Detects

1. Causal language without causal design -- "X causes Y" from correlational data
2. Universals without universal evidence -- "always", "never", "all cases"
3. Novelty inflation -- "first ever", "unprecedented", "never before"
4. Hedge removal -- conclusions stronger than the stated limitations allow
5. Scope creep -- claims that extend beyond the study population or conditions
6. Missing qualifiers -- assertions that lack "in this model", "under these conditions"

Example Output

LINE 142  STRONG  "This proves that serotonin drives mood"
          -> Suggest: "These results are consistent with serotonin contributing to mood"
          -> Reason: ODE model results do not constitute causal proof

LINE 287  MEDIUM  "For the first time, we show..."
          -> Suggest: "To our knowledge, this is the first demonstration of..."
          -> Reason: Cannot verify global novelty without exhaustive review

LINE 301  WEAK    "always produces oscillatory behavior"
          -> Suggest: "produces oscillatory behavior across all tested parameter ranges"
          -> Reason: Universal claim limited to explored parameter space

Exit Codes

• 0 -- no overclaims detected
• 1 -- overclaims found (see report)