toolclub/agent_team_gui
Reusable agent squads with per-agent provider/model routes and tool policies, serial/parallel dispatch, spawn/fork/chain context modes, and a Web management panel.
Install
via GitHub · GitHub source
Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.
What happened when we ran it
Installs, but needs a build approved first
pnpm refuses to run a package's build script until it is allowlisted. Add the key pnpm prints to allowBuilds in the profile's pnpm-workspace.yaml, then run the command again.
Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run 15d ago.
Show it in your README
For maintainers: the badge serves this listing's latest sandbox verdict, so a re-run updates it on its own — and it links readers to the full result here.
Due diligence
What it is — Provides reusable agent squads with per-agent model and tool policies, plus a Web management panel for planning and dispatching.
Who it is for — Suitable when you need planner, implementer and reviewer roles with different models assigned to each. Not for users seeking a single-model setup without team coordination overhead.
Watch out — Real-world install requires manually allowing the GitHub build script to bypass pnpm restrictions. No other issues observed in sandbox testing.
The verdict — I would install it because the team-based dispatch and per-member routing deliver more control than a solo agent, provided you handle the one-time build allowance.
Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.
agent_team_gui is a DeepSeek Harness plugin that runs reusable teams of agents with separate model routes, role prompts, fallback routes, token limits, and tool policies.
DeepSeek Harness loads the plugin in the Web profile and adds team controls beside the normal composer. A selected team responds to eligible top-level messages according to its activation and conversation mode. With fixed order disabled, a bounded, tool-free planner uses the active conversation's provider/model route, member roles, and the current request to produce structured assignments and an acyclic dependency graph. Ready members run in waves up to maxConcurrency; dependent members receive bounded structured handoffs, while full outputs, stages, retries, errors, timings, review rounds, and official token coverage are kept in durable run history. A fixed member order uses serial dispatch instead. Teams and members are managed in Settings → Members and Settings → Teams, and configuration can be checked with dsh --profile web --dump-config.
This is intended for developers who need repeatable multi-agent workflows across conversations and projects, with per-member permissions and observable runs. It is a poor fit for a single-agent task, or for environments without the Web profile, a supported Node.js version, and at least one configured provider/model route. The plugin also adds a terminal surface: any terminal-capable tool granted to a team member can be reached during that member's run, so member tool policies determine the resulting access.
DeepSeek Harness loads agent_team_gui in the web profile. Members are created in Settings → Members with a provider/model route, role prompt, optional fallback route, token limit, and selected tools. Teams are created in Settings → Teams from those members. Team settings include activation (Always, Smart, or Manual), member selection (All members or Adaptive), foreground or background response, fixed ordering, and optional resilience, budget, and review controls. The README names maxConcurrency as the limit for concurrently runnable DAG nodes.
Conversation selection supports Team, Solo, and Inherited. A team can be queued for the Next message, while a project default can supply the inherited selection. These states survive refresh, cold Host startup, and live reconnect.
DeepSeek Harness starts the Web profile with:
dsh --profile web
To inspect the composed bundle without installing rg, run:
dsh --profile web --dump-config | grep -E "agent-team-gui|dsh-agent-team-gui"
The output should contain the dsh-agent-team-gui bundle layer and the agent-team-gui row. If dsh is not on PATH, the README documents the equivalent pnpm dsh --version form from a Harness checkout and a pnpm --dir /absolute/path/to/deepseek-harness dsh ... prefix for commands.
DeepSeek Harness must be >=0.1.0-rc.5 <0.2.0, with the Web profile enabled. Supported Node.js versions are >=22.19.0 <23 or >=24.0.0; Node.js 23 is unsupported. The README also requires pnpm and at least one configured DSH provider/model route.
Git dependencies run the repository's prepare build. With pnpm 10 or later, that build may require allowBuilds: dsh-agent-team-gui: true in ~/.dsh/profiles/web/pnpm-workspace.yaml. Restart an existing DeepSeek Harness Web process after installation or an update.
Without a fixed order, the planner returns assignments and an acyclic graph; invalid, cyclic, or unavailable plans fall back to deterministic role-scoped assignments. Dependants receive bounded structured handoffs, not the complete preceding output. An optional quality gate can request up to two repairs from a named repair owner. Team runs expose plans, stages, outputs, errors, timings, retries, review or repair rounds, and official token coverage.
Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub
Same category
Q00
Config-only bundle that mounts Ouroboros through the DSH MCP client, exposing 36 interview, Seed, execution, evaluation, and evolution workflow tools in DSH.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
loopx-project
LoopX, a provider-neutral, local-first state kernel and control plane for long-horizon agents: keeps Goal, Todo, gate, evidence, quota, recovery, and handoff state above DeepSeek Harness, while the plugin bootstraps the CLI and skills, admits bounded same-session continuation, and adds a loopback GoalBar for the exact bound loop.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
huangruiteng
LoopX, a provider-neutral, local-first state kernel and control plane for long-horizon agents: keeps Goal, Todo, gate, evidence, quota, recovery, and handoff state above DeepSeek Harness, while the plugin bootstraps the CLI and skills, admits bounded same-session continuation, and adds a loopback GoalBar for the exact bound loop.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
chuspeeism
Embeds the active installed Codex Taskboard runtime in the DeepSeek Harness sidebar, using its launcher runtime descriptor instead of a fixed port.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
tong-io
TongFlow film-crew studio for image, voice, music and video production: the agent writes per-asset TongFlow workflow files (.tongflow.json) that run through TongFlow plugins, with an embedded workflow canvas, a shot/character/take project layout and a manga-drama template; sessions starting with @tongflow open the Studio view.
No one-line install — this plugin lives inside a larger repository and publishes no npm package.
