Why it matters
aSPARK offers AI builders a structured workflow to manage complex development cycles with AI coding assistants. By enforcing quality gates and maintaining a clear audit trail, it helps ensure features meet defined criteria and reduces the risk of scope creep. This can lead to more predictable and higher-quality AI-powered product development.

What changed

The aSPARK project, a Claude Code plugin, implements a gated agile delivery process for AI product teams. This process, detailed in its GitHub repository, structures development into five distinct phases: Specify, Plan, Act, Review, and Keep. Each phase must pass a quality gate before the next can begin, ensuring a rigorous progression from idea to release. The system generates a reviewable paper trail within a .spark/ directory, documenting each step of the feature development lifecycle, including specifications, architectural plans, code reviews, and QA reports.

Key roles are defined within the process, including Product Owner, Designer, Engineering Manager, Reviewer, QA Tester, and Release Manager, each associated with specific commands like /story-time, /sprint-plan, /increment, /peer-review, /demo-day, and /go-live. The orchestrator command /spark manages the entire loop, pausing at each gate for user approval. A "light path" is available for smaller changes that meet specific criteria, bypassing some review and QA steps to expedite delivery.

Recent updates, as of version v0.17.0, include changes to model usage for certain roles. Previously, the Engineering Manager and Reviewer were pinned to opus, but this has been updated. The Developer role, however, continues to run on the user's chosen model. The system also enforces cost and honesty controls, such as limiting Reviewer rounds to two by default and preventing code commits before an approved plan is in place, with aspark-guard 0.3.0 and later enforcing this denial.

Why it matters for builders

For AI builders, aSPARK provides a robust framework for managing the development of AI-powered products, particularly when leveraging tools like Claude Code. It addresses common challenges in AI development, such as maintaining quality, ensuring traceability, and managing complex workflows. The structured approach helps teams define clear acceptance criteria, track progress through a transparent paper trail, and integrate quality assurance, including real-browser QA, directly into the development loop. This can significantly improve the predictability and reliability of AI feature delivery.

The "light path" feature offers flexibility for simpler tasks, allowing teams to iterate faster without compromising the overall rigor of the process. By clearly defining roles and responsibilities and automating much of the workflow management, aSPARK aims to reduce overhead and allow developers to focus more on innovation and less on process management.

Practical impact

Developers can integrate aSPARK into their projects by installing it as a Claude Code plugin using /plugin marketplace add a-lottes/aSPARK and /plugin install aspark@aspark. For new projects, starting with /charter is recommended to define project principles. For existing codebases, the system can read the repository to establish initial documentation. Builders can explore complete examples like "steamcore" (an ESP32 arcade console) and the aSPARK repository itself, which uses aSPARK for its own development, to understand the .spark/ trail in practice. The system's commands facilitate each stage, from ideation (/story-time) to release (/go-live), with /spark orchestrating the entire sequence.

Caveats and source limits

The source material describes a specific process and toolset for Claude Code users. The effectiveness and scalability of aSPARK beyond the described use cases are not detailed. While the process emphasizes real-browser QA, the specifics of its integration and limitations for non-browser-based AI applications are mentioned as having declared substitute QA paths, but these are not elaborated upon. The exact impact of the model changes for the Engineering Manager and Reviewer roles on review depth or cost is not quantified. The "light path" criteria are described, but their real-world application and potential for misuse or misclassification are not explored. The source does not provide independent benchmark results for aSPARK's efficiency gains or quality improvements compared to traditional development workflows.

Sources

Written with AI assistance from the linked sources; every claim below was checked against them automatically. How we produce articles.

Claim check: 9/9 supported claims - 9 evidence links - 100% avg confidence
  • aSPARK is a gated agile delivery process for Claude Code, structured for AI product teams.supported - github.com
  • The aSPARK process includes five phases: Specify, Plan, Act, Review, and Keep, with quality gates between each.supported - github.com
  • aSPARK generates a reviewable paper trail in a `.spark/` directory.supported - github.com
  • aSPARK includes roles such as Product Owner, Designer, Engineering Manager, Reviewer, QA Tester, and Release Manager.supported - github.com
  • The latest release is v0.17.0.supported - github.com
  • In version v0.17.0, the Engineering Manager and Reviewer roles no longer run exclusively on 'opus' models, unlike previous versions.supported - github.com
  • aSPARK enforces controls such as limiting Reviewer rounds to two by default and preventing code commits before an approved plan.supported - github.com
  • aSPARK can be installed as a Claude Code plugin.supported - github.com
  • aSPARK offers a 'light path' for smaller changes that meet specific criteria, bypassing some review and QA steps.supported - github.com

Caveats

  • Single-source caution: verify critical details at the linked source.
Radar score 86/100 - how it was calculated
Reliability82
Freshness92
Novelty77
Technical81
Developer96
Ecosystem72
Confidence100
  • Reliability 82: GitHub metadata supports source trust
  • Freshness 92: Fresh GitHub activity
  • Novelty 77: Fresh GitHub release
  • Technical 81: Repository technical metadata
  • Developer 96: Developer tooling signals
  • Ecosystem 72: Fresh GitHub release
  • Confidence 100: Claims have reliable evidence
Share
XLinkedInHacker News

Discussion

Loading comments...

Related articles

AI Coding - Nov 13, 2026Gentle-AI v4.4.0 Enhances AI Coding Agents with Memory and WorkflowGentle-AI has released version 4.4.0, an open-source deterministic engineering environment designed to enhance existing AI coding agents. This update focuses on providing persistent memory, structured workflows, and evidence generation for agents like Claude Code, Cursor, and Codex, aiming to improve their reliability and context retention.AI Coding - Nov 19, 2026Termaxa: AI Agent Shell Command Safety GateTermaxa is a new Rust binary that acts as a gatekeeper for AI coding agents interacting with the shell. It previews command consequences, takes backups, and enforces policies before execution, aiming to prevent accidental damage from agents like Claude Code, Codex, Cursor, and Copilot.AI Coding - Nov 16, 2026Open Walnut: Self-Hosted AI Agent Workspace with Claude Code IntegrationOpen Walnut is a new self-hosted, local-first application that integrates AI agents, including Claude Code, into a single web interface. It offers durable coding sessions, task management, and note-taking capabilities, aiming to replace multiple developer tools.AI Coding - Nov 15, 2026OpenCode Mobile: Run AI Coding Agents on AndroidOpenCode Mobile is a new Android application that allows users to run various AI coding agents, including Claude Code and OpenCode, directly on their phones without needing a computer. The app provides a Linux environment, chat interface, diff viewer, and Git integration, enabling users to manage and steer agents remotely or locally.Agents - Nov 13, 2026Amplitude uses OpenAI models for product development agentAmplitude has developed a proactive product agent named Wave, which leverages OpenAI models. Wave assists in identifying product improvements, aids teams in shipping changes using coding agents, and measures the impact of these changes.AI Coding - Nov 20, 2026iPhone CLI: AI Agents Control Real iPhones via Accessibility TreeThe iphone-cli tool allows AI agents to interact with real iPhones using the accessibility tree, bypassing the need for Xcode, WebDriverAgent, or on-device app installations. This Rust-based CLI tool supports iOS 27+ and operates directly from the terminal on macOS and Windows.