> ## Documentation Index
> Fetch the complete documentation index at: https://mehen.ophi.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Link Debt

> Link health and reference burden for Markdown documents.

**Link Debt** quantifies link health as a 0–1 score. Broken and unresolved links dominate the score
because they are objective defects; external links are not bad by default but make a document fragile
when over-represented.

## Link classification

Every link in the AST is bucketed by destination:

* Internal anchor.
* Relative repository file.
* Absolute same-repo URL.
* External.
* Issue / PR.
* Bare URL.
* Image target.
* Broken / unresolved.

## Link Debt Score

```text theme={null}
broken_rate      = L_broken / max(1, L_total)
external_rate    = L_ext    / max(1, L_total)
bare_rate        = L_bare   / max(1, L_total)
anchor_miss_rate = missing_internal_anchors / max(1, L_int)

LinkDebtScore = clamp01(
    0.45 · sat(broken_rate;      0.00, 0.10)
  + 0.20 · sat(anchor_miss_rate; 0.00, 0.10)
  + 0.15 · sat(bare_rate;        0.05, 0.30)
  + 0.10 · sat(external_rate;    0.60, 0.90)
  + 0.10 · sat(link_density_per_100w; 6, 14)
)
```

## Companion scores

* **Information Scent Score** rewards descriptive link text, resolved relative links, working
  anchors, and a reference section when the doc is citation-heavy.
* **Link Review Burden** —
  `0.3·L_int + 0.8·L_rel + 1.0·L_ext + 2.5·L_broken + 0.5·L_footnote` —
  is the cost-per-PR signal used in diff reporting.

## Bands

| Link Debt   | Meaning  |
| ----------- | -------- |
| 0.00 – 0.20 | Healthy. |
| 0.21 – 0.40 | Mild.    |
| 0.41 – 0.60 | Inspect. |
| 0.61 – 0.80 | High.    |
| 0.81 – 1.00 | Severe.  |

## References

* Pirolli, P. & Card, S. (1999). *Information Foraging.* Psychological Review 106(4): 643–675 —
  origin of the "information scent" companion score.
  [DOI](https://doi.org/10.1037/0033-295X.106.4.643).
* Spool, J. M., Schroeder, W., Scanlon, T. & Snyder, C. (1998). *Web site usability: A designer's
  guide.* Morgan Kaufmann — empirical evidence for descriptive link text and resolved-anchor
  weighting.
* Nielsen, J. (1996). *Top ten mistakes in web design.* Nielsen Norman Group — broken-link
  prevalence as the dominant link-debt term.
  [Article](https://www.nngroup.com/articles/top-10-mistakes-web-design/).

## See also

* [DMI](/metrics/markdown/dmi) — uses Link Debt as one of its inputs.
* [Repository Grounding](/metrics/markdown/repository-grounding) — uses resolved link counts.