# Overclaim

> An ontology of epistemic-overclaim failure modes: named codes for the ways confident language outruns its evidence, in LLM and human prose. Overclaim is an agent-facing domain , it serves a rubric a model fetches and applies to a piece of text, judging whether confidence is earned rather than whether a claim is true.

To use: pick the rubric matching the text's genre, paste it into a context window above the subject text, and apply it. Severity is scored relative to the declared frame (genre/audience/goal), so an overclaim in an analytical memo is a defect while the same line in sales copy may be legitimate framing.

## Rubrics (paste-ready, per frame)
- [analytical](https://overclaim.org/rubric/analytical.md): factual/analytical writing , memos, reports, design docs, reviews
- [advocacy](https://overclaim.org/rubric/advocacy.md): advocacy , making a case, proposals, opinion pieces
- [sales](https://overclaim.org/rubric/sales.md): sales copy , pitches, landing pages, outbound
- [status](https://overclaim.org/rubric/status.md): status reporting , standups, updates, exec replies
- [persuasion](https://overclaim.org/rubric/persuasion.md): general persuasion , essays, talks, internal selling

## Packs (source of truth, raw YAML)
- [core](https://overclaim.org/packs/core.yaml): epistemic-overclaim core (evidence, narrative, communication, risk)
- [substantiation](https://overclaim.org/packs/substantiation.yaml): shared proof-bar axis (the FTC competent-and-reliable-evidence duty)
- [grounding](https://overclaim.org/packs/grounding.yaml): reasoning-environment warning lights (CC0, WillowEmberly / Negentropy)
- [management](https://overclaim.org/packs/management.yaml): commitment-without-mechanism and management-specific modes

## Docs
- [Code catalog](https://overclaim.org/codes/): every code with definition and examples
- [Design](https://overclaim.org/design/): why the ontology is the asset, the frame gate, prior art