By pedrogrande
Database-driven agentic pipeline for software development. Every agent action is authorised and scoped by a task record.
Implements features by reading specs, test specs, and DoD, writing code, then submitting a build report (step 7).
Writes the retrospective report and recommendations after tests pass (step 9). Produces DB records only — no file access.
Reviews and approves (or rejects with feedback) product manager outputs: project records, feature sets, decision records, and final cycle verification.
Ingests project briefs, defines features and DoD, manages decisions and decision artefacts, and handles backlog promotion across pipeline cycles.
Reviews test specs (step 6) against the feature's definitions of done, validates they are testable and complete, then approves or rejects with specific feedback.
Show pending tasks for a TaskFlow agent role. Lists what work is waiting and which project/feature each task belongs to. Use this to find out what an agent needs to do next.
Show the current state of the TaskFlow pipeline for a project — which steps are done, in progress, pending, blocked, or rejected. Use this to get an overview of where the pipeline stands.
Guide for the test_reviewer agent when reviewing test specs in step 6. Auto-loaded when a step-6 task is in progress.
Guide for the tester agent when executing tests and recording results in step 8. Auto-loaded when a step-8 task is in progress.
Start a new TaskFlow project. Prompts for a project brief (file path or inline text), creates the project record, and seeds the pipeline for the product manager. Use this to kick off a new development cycle.
Admin access level
Server config contains admin-level keywords
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A VS Code agent plugin that implements a database-driven agentic pipeline for software development. Every agent action is authorised and scoped by a task record. The database is the single source of truth for pipeline state, permissions, and context.
brew install uv or curl -LsSf https://astral.sh/uv/install.sh | shAdd to your VS Code settings.json:
"chat.pluginLocations": [
"/path/to/taskflow"
]
Or install from GitHub by adding the repo URL to chat.pluginLocations.
On first use, uv will automatically install the mcp dependency. No manual pip install required.
There are two ways to kick off a project. Use Path A when a client has filled in the brief form; use Path B for quick or informal starts.
Open docs/project-brief-form.html in any browser (no server needed — it runs offline).
Complete all sections: identity, goals, features, workflows, NFRs, integrations, risks, timeline.
Click Generate brief — a project-brief-<name>.json file downloads to your machine.
In VS Code Copilot chat, invoke the TaskFlow Product Manager agent and run:
@TaskFlow Product Manager use ingest_brief to start this project
Paste the contents of the JSON file (or provide the file path). The agent calls ingest_brief, which parses all structured data into the database and seeds the first pipeline task.
Proceed to Working the pipeline.
/start-project in Copilot chat and follow the prompts — paste a brief as plain text or provide a file path./my-tasks to see what each agent needs to do next./pipeline-status at any time to see the full pipeline state.| Command | Description |
|---|---|
/start-project | Start a project from free-text or a brief file (Path B) |
/my-tasks | Show pending tasks for a chosen agent role |
/pipeline-status | Show the full pipeline state for a project |
For the form-based path, use ingest_brief directly via the TaskFlow Product Manager agent rather than /start-project.
| Agent | Role in pipeline |
|---|---|
| TaskFlow Product Manager | Defines features, decisions, and decision artefacts (steps 3, 10, 12) |
| TaskFlow PM Reviewer | Reviews and approves PM outputs (steps 2, 4, 11, 13) |
| TaskFlow Tester | Writes test specs and runs tests (steps 5, 8) |
| TaskFlow Test Reviewer | Reviews test specs (step 6) |
| TaskFlow Builder | Implements features (step 7) |
| TaskFlow Documenter | Writes retrospective and recommendations (step 9) |
The brief form (docs/project-brief-form.html) is a single offline HTML file — no install, no server, no dependencies.
Sections captured:
| Section | What agents use it for |
|---|---|
| Project identity & problem | All agents — project scope and context |
| Goals & success metrics | PM (step 3 feature alignment); PM Reviewer (step 13 final verification) |
| User roles & workflows | PM (step 3 user-centric features); Tester (step 5 test scenario design) |
| Features (Must / Should / Could) | PM (step 3 starting point — promote Must features first) |
| Non-functional requirements | Builder (step 7 implementation constraints); Tester (step 8 verification) |
| Integrations | Builder (step 7 — system, direction, auth method, phase 1 flag) |
| Risks | PM (steps 10/12 — seeded as initial decision artefacts) |
| Release phases | PM (step 3 — assigns features to cycles) |
| Timeline & deadline | All reviewers — context for prioritisation |
Form features:
localStorage every 2 seconds — reload the page and it offers to restore the draftproject-brief-<slug>.json — the file is the pipeline entry artefactnpx claudepluginhub pedrogrande/taskflowSQLite-first developer workflow database with task-scoped agents, MCP tools, and slash commands
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Intelligent draw.io diagramming plugin with AI-powered diagram generation, multi-platform embedding (GitHub, Confluence, Azure DevOps, Notion, Teams, Harness), conditional formatting, live data binding, and MCP server integration for programmatic diagram creation and management.
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Harness-native ECC plugin for engineering teams - 67 agents, 277 skills, 93 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
Binary reverse engineering, malware analysis, firmware security, and software protection research for authorized security research, CTF competitions, and defensive security
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer