Blog tag
Tagged: developer.
20 articles tagged developer
Grok project management: give xAI's coding agent a real board
How to use Grok Build, xAI's coding agent, for real project management: connect it to a shared board over MCP, delegate tickets, and read its sessions.
Your coding agent re-fixes code that was already fixed
AI agents patch already fixed code in 35 to 65 percent of stale tickets. What a work item must carry so a fresh session tells done from open.
The agent's pull request is 80 percent right. Now what?
An AI agent's PR that is useful but not mergeable has three exits: finish it, send it back, or close it and rewrite the ticket. How to pick in five minutes.
Onboarding a new engineer to code your agents wrote
Git blame points at a merge and whoever wrote the prompt has left. How to onboard a new engineer to a codebase your AI agents wrote, without guessing.
What your junior engineer does now that agents take the tickets
A staffing plan for leads: what a junior developer works on when AI agents do the tickets, from spec authorship to first review of agent pull requests.
Did your AI agents actually save time, or did it just feel fast?
How to tell if your AI coding agents actually saved time: a four week check that runs on ticket data you already have, with a sign test and honest limits.
How to write a ticket an AI agent can actually finish
How to write tickets for AI coding agents: the seven fields that decide whether you get a reviewable pull request or an eleven file mess you cannot check.
What an AI agent actually costs per task
Work out AI agent cost per task: attribute agent-reported spend to an issue key, divide by completed work rather than attempts, and read the result honestly.
Shortcut alternatives for teams running AI agents
Shortcut alternatives for software teams: July 2026 pricing checked at the source, the seat math, and what changes when an AI agent is a real member.
Prompt injection starts in the ticket your agent reads
A stranger can write the ticket your AI agent reads. Real 2026 prompt injection incidents, what a hijacked agent can change, and how to cut the blast radius.
How to split work between Claude Code, Codex and Cursor
How to split work between Claude Code, Codex and Cursor: a division of labour that holds up, and where the shared state lives since none of them reads the others.
When a task file stops scaling for your AI agents
Beads is a good issue tracker for AI agents and a solo developer should stay on it. Here is the point a repo local backlog stops scaling, and what moving costs.
Turn a PRD into epics, stories and tasks agents can build
Give your AI coding agents a plan they can build. Paste a PRD, Taskfolk drafts the backlog, then Cursor, Claude Code and Codex work it through your workflow.
Claude Code task management: a real tracker, not a markdown file
Give Claude Code persistent, attributed task management by connecting a real tracker over MCP instead of a markdown file. Honest comparison to Task Master.
A shared backlog for a team of AI coding agents
Task manager for AI coding agents: give three agents one shared backlog with per-agent identity, field permissions, status handoffs, and human review gates.
Taskfolk integrations: the marketplace, OAuth apps, and the API
Taskfolk integrations explained: the 12-card marketplace, signed webhooks, scoped API keys, a first-party MCP server, and OAuth apps, plus what is missing and why.
Should your agent use MCP or the REST API?
MCP and a REST API solve overlapping problems for AI agents. Here is a plain guide to when each one is the right call, and why you often want both.
Why we gave Taskfolk an MCP server, and what it changes
What the Model Context Protocol is in plain terms, why a project tracker is a good fit for it, and how an agent picks up your workspace without any glue code.
AI agents on your board over the REST API: scopes and safe writes
A developer guide to giving an AI agent a scoped Taskfolk API key so it can move cards and file issues without a big blast radius.
Which project tools ship a first-party MCP server?
Having AI features is not the same as shipping an MCP server. Here is how to tell first-party from community-built from none, with an honest checklist.
