Plane adds agents to your workspace as members, with a scope and a playbook

What it does

Plane launched Plane Agents on Product Hunt on Tuesday, September 22, and it is live on Plane Cloud now. Plane calls itself work infrastructure for humans and agents: projects, knowledge and service on one work graph where people and AI agents operate as peers, in the company’s phrasing “structured enough for machines to reason over and humane enough for people to live in,” runnable on cloud, self-hosted, or fully air-gapped. More than 50,000 teams in 63 countries run it, many having switched from Jira, Asana, Linear and ClickUp; the company is certified to ISO 27001:2022, SOC 2 Type 2 and GDPR with a public trust centre, and the agents page lists Sony, Amazon, Accenture, Dolby, Aramco, Stark Bank, the République Française and the Government of Lithuania among its customers. The launch framing from Vamsi Kurama, who created Plane, is the whole idea in one sentence: “You can now add an agent to your Plane workspace the same way you’d add a team member. It gets a name, a set of projects it’s allowed to work in, and work assigned to it like anyone else.” An agent is assembled from six declared parts. A playbook — plain language covering what it is responsible for and how it should go about it, plus what a good result looks like and which boundaries it should follow. Skills you build yourself and attach, each a reusable way of doing a specific job, so any agent given one does that job the same way. Tools connected over MCP, letting it reach systems outside Plane when the job needs it. A scope — the list of projects it can act in and nowhere else. Triggers : assign it a work item, mention it in a thread, let a work-item change start it, or put it on a schedule, with filters deciding exactly when it runs. And memory , so it carries context between runs instead of starting cold and updates after each one. What it may do once hired: read and update work items, create new ones, comment, tag, move work between projects, and post summaries back to the team — triaging incoming requests, running a standup, drafting a spec from a rough note, watching for work that has stalled. “If a person can do it inside Plane, an agent can be given that job.” Five ready-made agents ship with it, each with declared skills and connectors: a Standup Agent that collects progress, blockers and next steps into one concise update; a Delivery Risk Agent watching for stalls, blockers, dependencies and slipping timelines across GitHub, Slack and your projects; a Spec Agent that turns rough ideas and requests into structured requirements, identifies gaps, and asks the questions that make scope clear; a Request Triage Agent that reviews incoming work, finds duplicates, adds context and routes it; and a Customer Feedback Agent that groups feedback into themes and connects them to the relevant issues and projects. The control section is where the launch earns attention. You choose each agent’s reach — which Plane projects, which trusted work sources, which connected tools. You choose whose access it uses : a person’s account, or a service account managed by the team. And you can see where your credits go, with adoption, response and completion rate, average time to complete, and consumption reported per agent. Every action an agent takes lands in the same activity history as your team’s, so there is no separate log to remember to check. On packaging, agents run on the AI credits already included in paid plans — no separate subscription and no per-agent seat — with a free tier to experiment on.