docs: Team Server Mode — running a fleet of concurrent agents on one shared machine#2313
Open
shulam-pay wants to merge 1 commit into
Open
docs: Team Server Mode — running a fleet of concurrent agents on one shared machine#2313shulam-pay wants to merge 1 commit into
shulam-pay wants to merge 1 commit into
Conversation
…ne shared machine Documents the third topology beyond default and team mode: N agents sharing a single server (devbox, Codespace, CI pool). Seven rules — per-agent identity, worktree isolation, sanctioned writers, git-backed brain, ship-lane locks, scoped per-command credentials, headless block-don't-guess — plus a threat model and adoption checklist. Extracted from a production multi-agent deployment; everything works with gstack as shipped, and the doc is structured so a future gstack-server-init could automate it.
|
Merging to
After your PR is submitted to the merge queue, this comment will be automatically updated with its status. If the PR fails, failure details will also be posted here |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
A pattern doc for the topology gstack doesn't cover yet: many concurrent AI agents sharing one machine (cloud devbox, Codespace, CI runner pool). Default mode is 1 human/1 laptop; team mode is N humans/N laptops/1 repo; this is N agents/1 machine.
Why
We run gstack-style multi-agent fleets in production at Skoor (x402 payments platform) — several Claude Code sessions working one shared checkout 24/7. Everything in this doc came from real collisions: branch switches under another agent's feet, unattributable checkpoint commits, double-ship races, and a shared-memory secret near-miss.
The pattern
Seven rules, all implementable with gstack as shipped today: per-agent identity (
GSTACK_AGENT_ID), worktree isolation (no branch switching in the shared checkout), one sanctioned writer per shared artifact, git-backed brain as the coordination substrate (memory sync promoted from optional to load-bearing), a flock-serialized ship lane, scoped per-command credentials +redact_prepush_hook, and headless sessions that block instead of self-approving. Includes a threat model and an adoption checklist, structured so a futuregstack-server-initcould automate it.Docs only — one new file + one README docs-table row. Happy to iterate on scope/placement, or to follow up with the
gstack-server-initscript if the pattern lands.