Requel

the Collaborative Requirements Elicitation System

Released Requel 2.0.1 is out. Get the jars from Releases or run the rreganjr/requel image; see Installing and running.

What Requel is

Poor requirements have been cited as the primary cause of software project failures and cost overruns. Frederick Brooks tells us that “The hardest single part of building a software system is deciding precisely what to build.”

What causes poor quality requirements, and why is it so hard to acquire requirements that are understandable, unambiguous, precise, and complete? One of the key problems in identifying requirements is a lack of stakeholder involvement. All stakeholders — including users, managers, and developers — need to be involved during requirements acquisition. Another problem is the high level of detail necessary to implement software. Most people don’t think about the intricacies of the tasks they perform at a conscious level.

Requel is a web-based requirements engineering system that supports collaboration among business and technical stakeholders and provides automated assistance to validate requirements and suggest improvements. It models requirements as goals, stories, actors, scenarios, and use-cases, with an IBIS-style annotation and discussion layer for raising issues, proposing solutions, and tracking decisions. It applies natural language processing to the requirements to detect and report ambiguity and complexity, and identifies potentially significant terms to assist in building a glossary and making the requirements more understandable and consistent.

What’s new in 2.0

  • A new web UI. The Echo2 server-side UI is replaced by an Angular single-page app, bundled into the same jar, with live refresh when other people or the assistants change something.
  • An API. Everything the UI does goes through POST /api/commands/{type} and GET /api/…, with personal access tokens for scripts.
  • AI requirements review. Assistants review goals, stories, use cases and scenarios and raise issues you can discuss and resolve. Use OpenAI, Anthropic, Gemini or a local model (Ollama); projects can write their own review definitions.
  • MCP server and requel-cli. AI clients (Claude, VS Code and others) can read and edit a project over MCP with OAuth sign-in; the CLI does the same from a terminal.
  • Stakeholder permissions that hang together. Permissions granted together are shown together, deletes that remove owned items are explained, and granting needs Grant.
  • Also: tags, report templates, project delete, links from requirements to their sources (tickets, documents), actors on stories, and an audit log of every command.

User guide

How to use Requel 2.0, with screenshots of the sample project.

Project links

Project file schema

Requel projects can be exported to and imported from XML. Exported files carry an xsi:schemaLocation pointing at the published schema below, so an export can be validated against the same definition the build uses.

An example project file that can be imported: Requel.xml.

Background

Requel began as a master’s thesis project at the Harvard Extension School (ALM, 2009). The thesis covers the background on requirements engineering and why it matters. The 2009 user guide describes Requel 1.0; for 2.0 see the user guide above.