Inspiration

Modern software teams rely heavily on Jira, but workflows tend to evolve organically. Over time, teams add statuses, automation rules, and shortcuts to move faster — rarely stopping to evaluate whether the workflow itself is still healthy.

We noticed a recurring pattern: teams only realize workflow problems after sprint goals are missed. Jira provides historical reporting, but it does not proactively detect structural risks like bottlenecks, dead ends, or conflicting automation.

We wanted to build something that acts as a guardrail for Jira workflows — continuously analyzing how work actually flows and surfacing risks early, directly inside Jira, without requiring configuration or behavior changes.

That idea became FlowSentry.


What it does

FlowSentry is a Jira-native workflow intelligence platform that proactively detects delivery risks inside Jira issues.

For every issue and workflow, FlowSentry:

  • Analyzes workflow structure to detect cycles, dead ends, and inefficient paths
  • Measures time spent in each workflow state to identify bottlenecks
  • Detects conflicts between Jira automation rules targeting the same fields
  • Computes a deterministic delivery risk score
  • Generates clear, actionable recommendations instead of raw metrics

All insights appear directly inside Jira using Issue Panels and Issue Glances. There are no external dashboards, no workflow changes, and no configuration required.


How we built it

FlowSentry is built entirely on Atlassian Forge, using a modular and deterministic architecture designed for production reliability.

Core Architecture:

  • Event Ingestion Layer Captures issue transitions and workflow metadata from Jira events.

  • Workflow Graph Analyzer Models workflows as graphs to detect cycles, dead ends, and unreachable states.

  • Timing & Bottleneck Analyzer Measures time spent in each workflow state to identify congestion and delays.

  • Automation Conflict Analyzer Detects overlapping or conflicting automation rules.

  • Risk Scoring Engine Aggregates multiple signals into a normalized, explainable risk score.

  • Jira UI Integration (Forge) Renders insights directly inside Issue Panels and Issue Glances.

All analysis is deterministic and explainable. We intentionally avoided black-box AI so teams understand why a risk exists, not just that it exists.

For hackathon evaluation, we added a clearly labeled Demo Mode that produces realistic results without requiring historical Jira data.


Challenges we ran into

  • Forge constraints and packaging Working within Forge’s strict module, resolver, and UI boundaries required careful architectural planning.

  • Balancing realism with demo usability We needed meaningful insights without relying on weeks of historical Jira data.

  • Rendering inside Jira consistently Ensuring stable UI behavior across different Jira views and environments.

  • Explainable risk modeling Designing a scoring system that is both useful and transparent, not a black box.

Each challenge pushed the architecture toward greater clarity and robustness.


Accomplishments that we're proud of

  • Built a fully native Forge app with real Jira UI integration
  • Designed a deterministic, explainable workflow risk model
  • Delivered actionable recommendations instead of vanity metrics
  • Achieved zero-configuration onboarding
  • Successfully deployed and validated inside a real Jira environment

Most importantly, FlowSentry solves a real, recurring problem faced by software teams every day.


What we learned

  • Workflow health is a systems problem, not just a reporting problem
  • Explainability matters more than opaque AI predictions in enterprise tools
  • Forge enables deep, secure Jira integrations when used correctly
  • The best DevOps insights live inside the tools teams already use

What’s next for FlowSentry

Post-hackathon, FlowSentry can evolve into a full workflow intelligence platform:

  • Sprint-level risk trend analysis
  • Organization-wide workflow health dashboards
  • Confluence auto-generated workflow reports
  • Predictive SLA breach detection
  • Optional AI-assisted explanations layered on top of deterministic logic

Our long-term goal is simple: help teams fix workflow problems before they become delivery failures.


App Information

App ID: f0631c69-d671-47ec-bebe-f607baf1a3a4


Share this project:

Updates