-
-
User Journey Diagram
-
Landing Page
Inspiration
Manual contract review is a common friction point for business teams, often leading to delays and compliance risks. We wanted to build a native solution that not only simplifies drafting but also introduces intelligent auditing that can distinguish between a valid signature and unauthorized tampering.
What it does
ContractGuard AI is a contract management suite built for the Atlassian ecosystem:
Drafting: Generates professional contracts based on user input for the contract type.
Smart Audit: Compares the signed document against the original to detect changes in legal clauses while ignoring harmless additions like signatures.
Storage: Keeps a record of the original draft within Forge's secure storage to ensure a reliable "source of truth".
Conversational Interface: By integrating with Atlassian Rovo, we’ve enabled teams to interact with their contracts through a custom Agent, making legal verification as simple as asking a question in natural language.
How we built it
Platform: Built on Atlassian Forge using the Node.js 22 runtime and the Latest UI Kit.
AI Engine: Powered by Gemini 2.5 Flash-Lite to provide high-speed legal reasoning and analysis.
Product Integration: We utilized the rovo:agent and action modules to extend our app’s functionality beyond the Jira issue panel, allowing it to act as a skill within the Rovo chat interface.
Security: All API credentials are encrypted using Forge Environment Variables to ensure enterprise-grade safety.
Challenges we ran into
A major technical challenge was moving from basic text-matching to AI-driven intent analysis. We had to refine our backend logic to ensure the system didn't flag signatures as "tampering." Additionally, configuring the Action modules in the manifest required precise mapping to ensure the Rovo Agent could accurately trigger our auditing functions
Accomplishments that we're proud of
Native Integration: Successfully built a tool that feels like a natural part of the Jira issue view.
Smart Comparison: Developed a reliable way for AI to distinguish between "tampering" and "signing".
Modern Tech Stack: Implemented the latest Gemini 2.5 models within the Forge ecosystem.
What we learned
We learned the importance of Forge's security model, specifically how to handle data egress safely through manifest permissions. We also gained a better understanding of managing state between Forge's frontend and backend resolvers.
What's next for ContractGuard AI
We aim to add multi-document support and automated "red-line" suggestions for incoming contracts. We also plan to add support for cross-product search within Rovo, allowing the agent to audit contracts stored in Confluence or Bitbucket as well.
Log in or sign up for Devpost to join the conversation.