---
name: watch-pages-with-fullsy
description: Re-capture web pages with Fullsy and determine whether they changed since a prior capture. Use for pricing pages, terms, policies, product claims, competitor pages, or recurring web-change checks where before-and-after evidence matters. Do not claim an exact textual diff unless the prior capture contents are actually available to compare.
---

# Watch pages with Fullsy

Re-capture important pages and report what changed, with before-and-after record ids as evidence.

## Preconditions

- Fullsy must be installed in the Chrome profile being automated.
- Fullsy Agent Access must be enabled and the control-page origin approved.
- You must be able to evaluate JavaScript in that same Chrome profile.
- For an exact content diff, the prior capture's contents must be available locally (for example from a downloaded Fullsy case) or otherwise readable. The bridge metadata alone can prove hash equality/inequality but cannot reconstruct old text.

If these requirements are not met, state the limitation rather than pretending to have compared evidence.

## Fullsy bridge

```js
await window.fullsy.captureUrl(url, { caseName })
await window.fullsy.getRecord(capId)
await window.fullsy.list(50)
```

Serialize captures: Fullsy allows one agent capture every 15 seconds.

## Workflow

1. Identify the exact page URLs to watch and the comparison point the user means by "last time".
2. Use `window.fullsy.list(50)` to locate the most recent prior agent capture for the same URL when available.
3. Fetch the prior record with `getRecord(priorId)` and retain its integrity manifest.
4. Re-capture the current URL into a clear case such as `Acme pricing watch 2026-08-17`.
5. Poll `getRecord(newId)` until the new record is stored.
6. Compare the old and new integrity manifests:
   - If relevant fingerprints are identical, report **no captured change detected**.
   - If fingerprints differ, report **the captured page changed**.
7. To explain *what* changed, compare actual prior and current page contents when both are available. Prefer semantic changes over noisy pixel/layout churn: price, plan, terms, claims, dates, availability, CTA, policy language, product scope, or other user-relevant differences.
8. If only manifests are available, do not invent a textual diff. Say that the fingerprints changed and that the prior content bundle is required to identify the exact delta.
9. Record both Fullsy ids for every comparison.

## Change classification

Classify confirmed changes when possible:

- `material` — changes meaning, price, obligation, eligibility, scope, or commercial claim
- `minor` — wording/design changes without meaningful semantic effect
- `unknown` — fingerprints changed but content-level evidence is unavailable
- `none` — relevant fingerprints match

## Final output

### Change summary
State the conclusion first.

### Before / after
| URL | Before record | After record | Result | Material change |
|---|---:|---:|---|---|

### What changed
Describe only differences actually supported by before-and-after content.

### Evidence limitations
Explicitly call out missing prior contents, capture failures, dynamic-page noise, or uncertainty.