# GitHub Brings Agent Activity Tracking to Issues and Projects

- Link: https://thestackobserver.com/github-brings-agent-activity-tracking-to-issues-and-projects/
- Published: 2026-03-30T08:43:32-04:00
- Author: Stackxx

GitHub has released two closely related features that surface AI agent activity 
directly within project management workflows. Released March 26, 2026, these capabilities
bridge the gap between AI-assisted coding and traditional project tracking, directly
addressing the growing visibility gap that emerges when non-human agents increasingly
perform work that was previously done exclusively by human developers.

## The Invisible Agent Problem

As coding agents like GitHub Copilot, Claude Code, and OpenAI Codex become standard
parts of modern development toolkits, a project management challenge has emerged
that few anticipated. When an AI agent checks out an issue ticket and begins working
on it, writing code, running tests, and making adjustments, that activity often 
happens entirely within invisible tool sessions or isolated sandboxes. The human
team sees an issue marked as in progress but does not actually know by whom, or 
more accurately by what, the work is being performed.

This lack of visibility creates coordination challenges. Multiple agents might unknowingly
work on related issues, duplicating effort. Managers cannot accurately assess sprint
progress because they cannot distinguish between issues being actively worked by
agents versus those languishing in the backlog. And when an agent finishes but the
issue remains open, debugging why requires detective work.

## Agent Sessions in Issue Sidebars

The first feature adds detailed agent session information directly under assignees
in GitHub Issues. When a coding agent is assigned to work on an issue, its active
session appears directly beneath human assignees in the issue sidebar, displaying
several key pieces of information.

A live status indicator shows the current state: queued when waiting for resources,
working when actively processing, waiting for review when code has been submitted
and needs human review, or completed when the agent considers the task finished.
Perhaps most valuably, these session entries include click-to-access session logs
for transparency and debugging.

## Project Board Integration

The second feature extends agent visibility into GitHub Projects table and board
views at scale. Team members can now see at a glance which items in a project have
attached agent sessions, understand their current status, and track how agent-assisted
work is progressing across a larger body of work spanning multiple issues and initiatives.

Enabling this enhanced view is straightforward: open the View menu in any GitHub
Project and toggle on Show agent sessions. This adds a dedicated column in table
views or visual indicators on board cards depending on the current view type.

## Why Visibility Matters

These features collectively address several concrete operational concerns that teams
face when integrating AI agents into their workflows. Coordination becomes significantly
easier when multiple agents might potentially pick up related issues, as visibility
prevents wasteful duplicate work. Teams can see that an agent is already addressing
a particular issue and redirect their own efforts elsewhere.

Estimation improves as well. Agent session indicators make it immediately clear 
which issues are actively being worked on versus those merely waiting in the backlog.
This visibility improves sprint planning accuracy and reduces overly optimistic 
assumptions about delivery dates based on misunderstood capacity.

## Integration and Implications

Agent sessions appear alongside existing GitHub project features without replacing
or competing with them. An issue can simultaneously have both human and agent assignees,
reflecting the reality of hybrid development where humans and AI collaborate. Sessions
update automatically as the agent progresses through its work, requiring no manual
status synchronization or check-ins.

As agent activity becomes visible and normalized in project management surfaces,
teams may naturally reconsider how they assign work. Issues particularly well-suited
for AI agents, those that are well-defined, narrowly scoped, with clear acceptance
criteria, can be explicitly routed to agents while humans focus on more ambiguous
or creative work.

## Sources

[Agent activity in GitHub Issues and Projects – GitHub Changelog](https://github.blog/changelog/2026-03-26-agent-activity-in-github-issues-and-projects/)