diff --git a/apps/vc-api/docs/architecture/decisions/0002-use-credo-for-key-and-credential-operations.md b/apps/vc-api/docs/architecture/decisions/0002-use-credo-for-key-and-credential-operations.md
new file mode 100644
index 0000000..4a74450
--- /dev/null
+++ b/apps/vc-api/docs/architecture/decisions/0002-use-credo-for-key-and-credential-operations.md
@@ -0,0 +1,35 @@
+# 2. Use Credo for key and credential operations
+
+Date: 2024-09-25
+
+## Status
+
+Accepted
+
+## Context
+
+Prior to this decision, Spruce's DIDKit was used for DID generation and credential issuance & verification.
+The rational for DIDKit's use is that it:
+- Is written in Rust and so suitable for use in any mobile app development framework
+- Supports JSON-LD and JWT credential issuance and verification
+- Supports did:key, did:ethr, did:web
+- DIDKit (and its libraries) are open-source
+
+However, a significant limitation of DIDKit is that only inline JSON-LD contexts are supported 
+(see [this discussion on the Spruce Discord server](https://discord.com/channels/862419652286218251/1021707856401141760/1021709845931495444)), 
+which limits the usability of the VC API application.
+
+## Decision
+
+Credo (formerly Aries Framework Javascript) is an alternative library for key and credential operations.
+We have decided to replace DIDKit with Credo.
+
+## Consequences
+
+Expected benefits from using Credo are:
+- The ability to provide additional JSON-LD context documents to a document loader
+- Synergy with another Open Wallet Foundation project
+- Improved typing, bug patching and debugging due to Credo also being written in Typescript
+
+A possible risk is that Credo encapsulates the database access
+and so the project will have less control over the database schema used for key storage.