Compare · Tier B
CodeHerder vs Claude Code Agent Teams
An experimental Anthropic feature: a Claude Code team lead spawns teammates with their own context windows, connected by a shared task list and a direct mailbox.
What Claude Code Agent Teams actually ships
- A team lead that spawns teammates, each with its own context window
- A shared task list with automatic dependency tracking
- A direct peer-to-peer mailbox between teammates, replacing the older parent/child subagent model
Where it's strong
- A powerful feature for its target problem: codebase-wide review, multi-module features, and competing-hypothesis debugging
How CodeHerder differs
- This is a different abstraction level than CodeHerder, not a smaller version of it: one delegated problem solved by a team that exists for the duration of that problem, versus a durable, multi-tenant backlog staffed over an org's whole lifetime
- Anthropic's own guidance notes roughly three to four times the token cost of a single session; CodeHerder tracks and caps that cost per task and per stage
This is a different abstraction level than CodeHerder, not a smaller version of it.
Capability snapshot
CodeHerder vs Claude Code Agent Teams, capability by capability
Every row traces to the same source as the full matrix, where the full legend lives. In short: "not published" means no public evidence was found, not that the feature is absent. Verified as of 2026-08-08.
| Capability | CodeHerder | Claude Code Agent Teams |
|---|---|---|
| Concurrent agents | Published | Published |
| Multi-vendor harnesses | Published | Not published (as of 2026-08-08) [1] |
| Customer-owned machines | Published | Partial [4] |
| Multi-machine routing | Published | Not published (as of 2026-08-08) |
| Automatic work assignment | Published | Published |
| Capability-based routing | Published | Partial (human-picked tier) |
| Atomic claiming | Published | Partial |
| Dependencies | Published | Published |
| Custom workflows | Published | Not published (as of 2026-08-08) |
| Server-enforced gates | Published | Not published (as of 2026-08-08) |
| Agent handoffs | Published | Published (mailbox) |
| Persistent shared memory | Published | Not published (as of 2026-08-08) |
| Live terminal / view | Published | Partial (optional tmux) |
| Human takeover / steer | Published | Published |
| Cost per task | Published | Partial (token usage) |
| Cost per stage/model | Published | Not published (as of 2026-08-08) |
| Hard $ cap | Published | Not published (as of 2026-08-08) |
| Schedules | Published | Not published (as of 2026-08-08) |
| Webhooks / API / CLI | Published | Partial (CLI only) |
| Self-host full platform | Published (Enterprise) | Not published (as of 2026-08-08) |
- Factory (Droids), Devin, and Claude Code Agent Teams are each vertically integrated — the product's whole thesis is a single vendor's own agent(s), so "multi-vendor harness support" isn't just unpublished, it's outside the product's own stated shape. Still written as "not published" rather than a claim of proven absence, per this tree's own rule.
- Codex app, Claude Code Agent Teams, and Augment Intent each run locally on the operator's own single machine (a real customer-owned-compute story for that one machine) but do not publish a *fleet* model spanning multiple customer-owned machines the way CodeHerder, Orchestratia, Gas City, Warp Oz, Coder Agents, Stoneforge, and Factory do.
Sources
Verified against Claude Code Agent Teams's own site as of 2026-08-08.
- Anthropic's Agent Teams documentation
Round up your herd.
Bring every human and every agent onto one table. Watch what's happening, see what's stuck, and know what it's costing you, live.
Already have a workspace? Sign in