
Issue Cards
https://github.com/obra/issue-cards- Category
- Productivity
- Rank
- No. 1019Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- obra
- GitHub
- 12 stars
- Date
About
AI-optimized command line issue tracking tool, designed for agents to read and update tasks.
What it does
Issue Cards keeps project work as structured Markdown records, then guides execution one task at a time. Issues capture the problem, planned approach, questions, failed attempts, instructions, and ordered tasks. Humans use terminal commands, while coding assistants can work through its MCP interfaces.
Why it's ranked here
The tool has a coherent opinion: keep agents focused on one documented step, not an entire issue. Local Markdown storage, parseable output, task expansion, and direct assistant access reinforce that goal. The author’s caveat about limited real-project use warrants caution, but the workflow is concrete and inspectable.
What's good
Issue records preserve decisions that agents often lose, including unresolved questions and failed approaches. Tags expand testing requirements into consistent steps. JSON and quiet output support scripts. Git integration stages issue updates automatically, while schema validation rejects malformed MCP arguments before tool execution.
Tradeoffs
The workflow is intentionally linear, which may constrain projects that need parallel task execution. Automatic Git staging changes repository state during routine updates. The HTTP server can run without authentication, so remote exposure requires deliberate token configuration. Its value also depends on maintaining detailed issue records inside each project.
How to use it well
Use it for repository-local coding work where an agent should complete, document, and close one step before receiving the next. Start with a clear problem, approach, and ordered tasks, then record questions and failures as work proceeds. Choose another system for nonlinear planning or a hosted shared workspace.
Technical notes+
package.json defines a CommonJS JavaScript package for Node.js 14 or newer, with CLI, stdio MCP, and HTTP MCP binaries plus Jest, ESLint, and documentation-validation scripts. src/cli.js discovers command modules dynamically through Commander. src/index.js centralizes global output modes and exposes executeCommand for programmatic execution. src/mcp/tools.js wraps issue operations with workflow guidance, while src/mcp/validator.js uses Ajv schemas with coercion, defaults, and rejection of extra properties. src/mcp/mcpServer.js mounts JSON routes under /api and supports bearer or query-token authentication. src/utils/template.js renders issue and tag templates with Handlebars.
Observed
- License
- MIT
- Primary language
- JavaScript using CommonJS modules
- Runtime
- Node.js 14 or newer
- Installation
- Global npm package or on-demand npx execution
- Interfaces
- Command line, programmatic module execution, stdio MCP, and HTTP MCP API
- Storage
- Markdown issue files stored directly within the project
- Validation
- MCP arguments are validated with Ajv JSON schemas
- Testing surface
- Jest scripts cover unit, end-to-end, MCP, link, and coverage workflows
Read from README.md, package.json, src/cli.js, src/index.js, src/mcp/tools.js, src/utils/errors.js, src/utils/output.js, src/commands/help.js, src/commands/init.js, src/commands/list.js, src/commands/show.js, src/mcp/mcpServer.js, src/mcp/validator.js, src/commands/serve.js, src/utils/template.js.
What it can do
Create new issue tracking tasks
Task description and metadata → New issue record in tracking system
Read existing issue details
Issue ID or search criteria → Issue information and current status
Update issue status and properties
Issue ID and updated field values → Modified issue record
Search and filter issues
Search query or filter parameters → List of matching issues
Optimize task prioritization using AI
Issue list and context data → AI-recommended task priorities
Execute commands through CLI interface
Command line instructions → Command execution results
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.