# Review What Changed, explained

**Which parts of a protocol's audits still apply to the code running today? A plain account of the problem, a real example and what other tools already do.**

Explainer · 14 November 2026 · Plain-language companion to Initiative 2 of 3, TheDAO Security Fund Round 2.

Online: https://proofofaudits.com/explained/review-what-changed

## In short

Protocols keep shipping after their audit. Contracts are upgraded, modules are added, libraries are replaced. The audit reports stay as they were, and it's often unclear which parts of them still apply to the code running today.

Review What Changed builds open tools that tie each audit to the exact code it covered and show what an upgrade touched, so a team can see what needs another look before it ships.

## Live demos

- [Audit-gap map](https://proofofaudits.com/demos/audit-gaps)
- [Pendle example](https://proofofaudits.com/explorer/functions?p=pendle)
- [Function explorer](https://proofofaudits.com/explorer/functions)

## 1. The problem

An audit report names a commit and a list of files. After that, the code moves on: a new router, a new facet, an upgraded library. Each change may be reviewed by a different firm with a different scope, or not reviewed at all.

The protocol's audits page still lists every report, so from the outside the protocol looks covered. Working out which of today's deployed functions those reports actually cover means reading every report's scope section and matching it against the live code by hand. Few teams do it, and almost nobody publishes the result.

## 2. A real example

Pendle's audit folder holds 23 PDF reports. We searched all 23, and none names the files of RouterV4 or the seven action contracts behind it, which together hold 58 callable functions. ChainSecurity's 2024 core audit says that files it does not list are out of its scope, and its list has no router file ([see the Pendle functions](https://proofofaudits.com/explorer/functions?p=pendle), [the reports we read](https://proofofaudits.com/evidence/supplements/audit-search.v1.json)).

That doesn't prove the router was never reviewed. It means no public report shows that it was, and anyone relying on "Pendle is audited" can't tell.

Across ten large Ethereum protocols we mapped 656 callable functions: 584 sit in a file a public report names, and 72 have no matching report ([the full map](https://proofofaudits.com/demos/audit-gaps)).

## 3. What other tools already do

*Existing tools and the question each answers*

| Tool | What it does | The question it answers |
| --- | --- | --- |
| Fix reviews and diff audits by audit firms | A firm rechecks fixes or reviews the changes between two commits. | Is this specific change safe? (Paid per engagement, published as a PDF.) |
| Audit contests ([Sherlock](https://sherlock.xyz/post/what-is-sherlock-an-explainer-for-builders-and-teams), Code4rena, Cantina) | Many researchers review a frozen snapshot of the code. | What bugs are in this snapshot? |
| [Solodit by Cyfrin](https://docs.solodit.cyfrin.io/) | A searchable database of findings from many audit firms. | What has been found before? |
| [OpenZeppelin Monitor](https://docs.openzeppelin.com/monitor) | Watches onchain activity, such as upgrades, and sends alerts. | Did something just change onchain? |
| [Etherscan contract verification](https://docs.etherscan.io/contract-verification) | Confirms published source compiles to the deployed bytecode. | Is this the source of the deployed code? |

These are useful, and we rely on several of them. None keeps a public, checkable record of which audit covers which deployed function, and updates it when the code changes.

## 4. What Review What Changed adds

- **Audit-to-code links:** each report tied to the commit, files and deployed contracts it covered, with the report's hash and a quoted line for every scope claim.
- **Change reports:** after an upgrade, the functions that changed, the callers and dependencies they affect, and whether they were in an earlier audit's scope.
- **Fix-recheck records:** a finding reproduced, fixed and rechecked against the patched commit, with the test attached.
- **Open data:** the gap map kept current and refreshed when protocols upgrade, free for researchers, contest platforms and the protocols themselves.

Protocols still pay for their own reviews. The tools help them spend that money on the code that actually changed.

## 5. What it does not claim

- "In audit scope" means a report names the file. It does not prove every function in it was examined.
- A missing report is a gap in public evidence, not a vulnerability.
- We don't replace auditors or claim ownership of their work. Every report keeps its authors' names.

## References

- [Pendle functions in the explorer](https://proofofaudits.com/explorer/functions?p=pendle)
- [Every audit search we ran](https://proofofaudits.com/evidence/supplements/audit-search.v1.json)
- [Audit-gap map demo](https://proofofaudits.com/demos/audit-gaps)
- [Full whitepaper](https://proofofaudits.com/whitepapers/audit-gap-map)
- [Solodit](https://docs.solodit.cyfrin.io/)
- [OpenZeppelin Monitor](https://docs.openzeppelin.com/monitor)
- [Etherscan contract verification](https://docs.etherscan.io/contract-verification)
