Workspaces are becoming native runtimes for AI agents
A repository and package cohort tests whether persistent state, shared artifacts, permissions, handoffs, and completed actions are replacing chat-only AI features.
Early evidence; the market is not yet confirmed
197 subject-filtered implementations indicate that agent workspaces are becoming more than chat shells. Persistent state appears in 2.5% of the cohort, shared artifacts in 1.5%, handoffs in 10.7%, and completed actions in 0%, versus chat-only language in 2.5%. The runtime pattern is visible, but governance and durable collaboration remain less common than orchestration.
Market snapshot
Comparable measurements from independent market surfaces.
Implementations and packages
Deduplicated and subject-filtered primary cohort.
Median repository stars
Calculated across repositories with at least one star.
With persistent state
5 items in the subject-filtered cohort.
With completed actions
0 items in the subject-filtered cohort.
Search-match dynamics
Bars show monthly GitHub query matches for agent and multi-agent workspaces; loaded counts are the implementations retained after subject filtering.
What exists inside the category
One item may contain more than one feature.
Persistent workspace state
5 · 2.5%
Shared artifacts
3 · 1.5%
Permissions and approvals
4 · 2%
Handoffs and delegation
21 · 10.7%
Multi-agent collaboration
54 · 27.4%
Completed actions
0 · 0%
Chat-oriented interfaces
5 · 2.5%
Representative projects
Cross-source validation
These publications are not part of the primary numeric cohort.