diff --git a/harness/README.md b/harness/README.md
index 4eb5677..28a2bc6 100644
--- a/harness/README.md
+++ b/harness/README.md
@@ -99,7 +99,10 @@ round (`progress.md`, `subagents.md` §5–§13).
   `team_add` gains `parent`, `mode`, `request`; `team_send` takes `parent`; `team_stop` is the main
   agent's. `blocks/team_submanager.md`; `team_lead.md` and `team_member.md` reworded. Over the link,
   capability `tree`: the `team` event, `_lembas/team`, `/transcript` (read-only), `/findings`,
-  `/stop`, `subagent` on an approval. Areas `team` (cases changed where a member used to add or
+  `/stop`, `subagent` on an approval. **An agent chat's subagents are the device's** (`subagents.md`
+  §14, `acp.md` *Subagents in an agent chat*): capability `agents` carries the chat's `{ min, max }`
+  down on every turn (and `said`, the person's earlier words, on `session/new`); the tree, progress,
+  approvals, findings and transcripts come up. Areas `team` (cases changed where a member used to add or
   stop, and new ones) and `delegation`.
 
 ## 3.3.0
diff --git a/harness/acp.md b/harness/acp.md
index a1bd0f9..eead803 100644
--- a/harness/acp.md
+++ b/harness/acp.md
@@ -66,6 +66,7 @@ change to the protocol is a change here first, then in both projects.
 | `show` | spec 3.0.0: `_lembas/show`; the `show` tool is offered in a session the client opened when the client said `show: true` in `clientCapabilities._meta.lembas` |
 | `jobs` | background work (spec 2.0.0): the `job` and `job_note` events; `_lembas/jobs`, `_lembas/job/output`, `_lembas/job/stop`, `_lembas/job/send`, `_lembas/job/transcript`; `jobs` in `_lembas/session/status`; `job` on `session/request_permission`; `machine` on a `prompt` event |
 | `progress` | spec 3.4.0 (`progress.md`): the `phase` event; `progress` in `_lembas/session/status` |
+| `agents` | spec 3.4.0: the session's Subagents setting `{ min, max }` is read from `_meta.lembas.agents` on `session/new`, `session/prompt` and `_lembas/configure` (*Subagents in an agent chat*, below) |
 | `tree` | spec 3.4.0 (`subagents.md` §5–§13): the `team` event; `_lembas/team`, `_lembas/team/transcript`, `_lembas/team/findings`, `_lembas/team/stop`; `subagent` on `session/request_permission` |
 
 ## Standard ACP, as used here
@@ -254,6 +255,57 @@ Answer: ACP's `{ outcome: { outcome: "selected", optionId } | { outcome: "cancel
 optional `_meta.lembas.command` (the corrected command line, `once` only) or
 `_meta.lembas.feedback` (why it was denied). No answer within `remote.approval_timeout` is a denial.
 
+## Subagents in an agent chat (spec 3.4.0)
+
+**An agent chat's main agent is the CLI on the device.** Its subagents are the device's: it starts
+them locally, on the models of the server it is logged in to, under every rule of `subagents.md` —
+the model rules (§2–§3), Min/Max, slots, permissions only ever narrowed, request options, and a
+call nobody in the tree can allow going to the person. The server runs no subagent of its own for
+an agent chat and offers its own subagent tool to no agent chat; what it does is carry the setting
+down and the tree up. A device without the capabilities `agents` and `tree` gets neither and runs no
+subagents for the chat (the chat says so beside the setting).
+
+**What goes down, with every turn.**
+
+- `agents: { min, max }` in `_meta.lembas` of `session/new`, of **every** `session/prompt`, and of
+  `_lembas/configure` when the person changes it between turns — the chat's setting as the server
+  keeps it (`team.md` §2: 1 ≤ min ≤ max ≤ the instance's cap). The device takes the value as the
+  session's setting for that turn: a change reaches the main agent as `team.md` §10's note, and
+  nothing else reads the device's own configuration for the chat. Absent (an older server): `{ min:
+  1, max: 1 }` — no subagents. A value the device's own global cap is under is read as `team.md` §2
+  says (lowered, never refused).
+- `subagents: false` in the same `_meta.lembas` when the instance has subagents switched off: the
+  device offers none whatever `agents` says.
+- **What the person named** (`subagents.md` §2): the person's words reach the device as the turns'
+  prompt texts, so the device judges a named model from them as in its own sessions — every
+  `session/prompt` without `machine`; never a `machine` turn, a `_lembas/steer` from the harness, or
+  a tool's output. A chat continued on a device that has not seen its earlier turns gets them in
+  `_meta.lembas.said: string[]`, the person's earlier messages of the chat, oldest first (on
+  `session/new` only); the device counts them as typed in the session.
+- The models: the device's subagents may run on any model of the person's (`/v1/models` as the
+  device's token sees it), as in its own sessions.
+
+**What comes up.**
+
+- The tree, as it changes: the `team` event (`{ tree }`, `subagents.md` §12's data, at most four a
+  second), and on request `_lembas/team` (the tree and the whole room). Every agent's `progress`
+  is in it; the main agent's own also as the `phase` event (capability `progress`).
+- Approvals: a subagent's call that only the person can allow is an ordinary
+  `session/request_permission` with `_meta.lembas.subagent: { id, name, position }`; the server
+  shows it as its usual approval card naming the subagent, in every client. Requests the main agent
+  decides (`team_request` / `team_decide`) never reach the server as approvals: they are room
+  messages.
+- Findings and transcripts on request: `_lembas/team/findings`, `_lembas/team/transcript` (live and
+  ended subagents, read-only — the link has no method that writes into a subagent's session).
+- The person stopping a subagent: `_lembas/team/stop`.
+
+**What the server keeps.** The latest tree with the chat (so a client opening the chat draws it at
+once, and after the device has gone), and the room as `_lembas/team` last answered. Findings and
+transcripts stay on the device; a client asking for one while the device is not linked is told it
+is on the device, and gets it once the device links again. Over `/api/v2` the chat's stream carries
+the tree as it changes and the endpoints to read a subagent's transcript and findings, the same for
+a chat's own subagents and for an agent chat's — so Web, Desktop and Android draw both alike.
+
 ## `_lembas/*` methods
 
 Direction is who sends it. "→ result" marks a request; the others are notifications.
@@ -307,7 +359,7 @@ Only on the link (an error on `serve --stdio`).
 
 ### `_lembas/configure` — client → agent → result
 
-Params `{ sessionId, model?: <instance's name>, effort?: <level> }` between prompts (a level as in `session/new`). Result
+Params `{ sessionId, model?: <instance's name>, effort?: <level>, agents?: { min, max } }` between prompts (a level as in `session/new`; `agents` with capability `agents`, spec 3.4.0). Result
 `{ model: <ref>, effort }`. Emits a `model` event when something changed.
 
 ### `_lembas/steer` — client → agent → result
diff --git a/harness/conformance/team.json b/harness/conformance/team.json
index 0b2b060..ec53ead 100644
--- a/harness/conformance/team.json
+++ b/harness/conformance/team.json
@@ -1,5 +1,5 @@
 {
-  "about": "Teams (team.md, spec 3.2.0; the main agent and its tree, subagents.md §5–§13, spec 3.4.0 — a member's team_add is now the lead's with `parent`, and `offered` lists 3.4.0's tools): one lead session's team, step by step, written from a reference model of the rules. `start`: the size setting `min` and `max`, `subagents` (enabled for the session; default true) and the chat model's `capacity` (models/schema.json: `parallel`, `single_session`; {} — no limit). The lead is `lead`; its first prompt is running as the case begins, and every agent is on the chat's model, and it is the only model there is (spec 3.3.0: other models are subagents.json's). Steps and what each answers: `offered` (`by` an agent, default lead) — `{tools, setting}`: the team tools listed for that agent, sorted, and whether the setting is shown; `call` (`by`, `tool`, `args`) — `{result, delivered}`: the call's answer word for word (null where an ordinary tool would run: the case looks only at whether it is refused), and what was delivered to whom by this step, `{name: [message, …]}`; `team_read` answers `{room, delivered}` instead, `room` being the messages it shows as `{n, from, to}` (the tree and the text format are loop.json team.read's, checked by each project's own tests); `task` runs a subagent that counts toward the size (and holds a slot) until `task_done` (`by`) — answers null; `finish` (`by`, `reply`, `reason`: stop by default, or steps, budget, `error — <text>`) — the agent's turn ends: a member's reply is its report — `{delivered}`; `set` (`min`, `max`, `capacity`: the person's setting or the model's metadata changed) — null; `person_stop` (`name`) — the person stopped that member in the UI — null; `user` — the person's message to the lead — `{note}`: the note on it (context.md §10), null when there is none; `state` — the team in tree order as `{name, manager, state}`: working, waiting (for a slot), finished, idle (the lead between turns). Delivering to a finished member or the idle lead wakes it (it is then working, or waiting when its model is full). LLeMbas: services/team.py; CLI: team/.",
+  "about": "Teams (team.md, spec 3.2.0; the main agent and its tree, subagents.md §5–§13, spec 3.4.0 — a member's team_add is now the lead's with `parent`, and `offered` lists 3.4.0's tools): one lead session's team, step by step, written from a reference model of the rules. `start`: the size setting `min` and `max`, `subagents` (enabled for the session; default true) and the chat model's `capacity` (models/schema.json: `parallel`, `single_session`; {} — no limit). The lead is `lead`; its first prompt is running as the case begins, and every agent is on the chat's model, and it is the only model there is (spec 3.3.0: other models are subagents.json's). Steps and what each answers: `offered` (`by` an agent, default lead) — `{tools, setting}`: the team tools listed for that agent, sorted, and whether the setting is shown; `call` (`by`, `tool`, `args`) — `{result, delivered}`: the call's answer word for word (null where an ordinary tool would run: the case looks only at whether it is refused), and what was delivered to whom by this step, `{name: [message, …]}`; `team_read` answers `{room, delivered}` instead, `room` being the messages it shows as `{n, from, to}` (the tree and the text format are loop.json team.read's, checked by each project's own tests); `task` runs a subagent that counts toward the size (and holds a slot) until `task_done` (`by`) — answers null; `finish` (`by`, `reply`, `reason`: stop by default, or steps, budget, `error — <text>`) — the agent's turn ends: a member's reply is its report — `{delivered}`; `set` (`min`, `max`, `capacity`: the person's setting or the model's metadata changed) — null; `person_stop` (`name`) — the person stopped that member in the UI — null; `user` — the person's message to the lead — `{note}`: the note on it (context.md §10), null when there is none; `state` — the team in tree order as `{name, manager, state}`: working, waiting (for a slot), finished, idle (the lead between turns). Delivering to a finished member or the idle lead wakes it (it is then working, or waiting when its model is full). Spec 3.4.0 adds: `start.mode` — the main agent's mode (default edit); `call` to team_whoami, team_finish, team_findings, team_request, team_decide, team_assign, team_replace, answered word for word; `context` (`by`, `used`, `window`) — its engine's context use — null; `context_full` (`by`) — its context reached team.context.pause_at: `{delivered}`; `started` (`by`) — `{text}`: what its latest turn was started with (a brief, or the messages that woke it); `tree` — `{size, open_slots, agents: [{id, name, parent, state, mode, request}], ended: [{id, name, state}]}`; `main_mode` (`mode`) — the main agent's mode changed — null. LLeMbas: services/team.py; CLI: team/.",
   "cases": [
     {
       "about": "Offered on a model that serves two requests at once: the lead has all four tools and the setting is shown.",
@@ -753,7 +753,7 @@
       ]
     },
     {
-      "about": "A task subagent counts toward max while it runs, inside a team.",
+      "about": "A task subagent counts toward max while it runs, inside a team. A subagent's own task is refused: only the main agent starts subagents (spec 3.4.0).",
       "start": {
         "min": 1,
         "max": 3,
@@ -826,7 +826,7 @@
           "delivered": {}
         },
         {
-          "result": "The team is full: it has as many agents as the person allows (3). Give the work to a member already on it (team_send), or stop one (team_stop) first.",
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
           "delivered": {}
         },
         null,
@@ -3416,6 +3416,1543 @@
           }
         ]
       ]
+    },
+    {
+      "about": "Spec 3.4.0: only the main agent adds, assigns, replaces, stops and decides; a subagent's call to any of them, or to task, is refused with one text.",
+      "start": {
+        "min": 1,
+        "max": 4,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_assign",
+          "args": {
+            "name": "api",
+            "task": "x"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_replace",
+          "args": {
+            "name": "api"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_stop",
+          "args": {
+            "name": "all"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_decide",
+          "args": {
+            "id": "r1",
+            "allow": true
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "task",
+          "args": {
+            "prompt": "x"
+          }
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "result": "Only the main agent adds, assigns, replaces and stops subagents: ask it (team_send to lead).",
+          "delivered": {}
+        },
+        {
+          "size": 2,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: a subagent is placed under a sub-manager with parent; parent must be one (lead: true), and the size still counts the whole tree.",
+      "start": {
+        "min": 1,
+        "max": 6,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "api"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "auth",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "routes"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "auth",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "ghost"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "parent",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "call",
+          "by": "routes",
+          "tool": "team_send",
+          "args": {
+            "to": [
+              "parent"
+            ],
+            "message": "Routes are in."
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_send",
+          "args": {
+            "to": [
+              "parent"
+            ],
+            "message": "Hello?"
+          }
+        },
+        {
+          "op": "finish",
+          "by": "routes",
+          "reply": "Routes done."
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added routes (owns a part): it is working. Its report comes to you when it finishes. The team is 3 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "routes is not a sub-manager: place a subagent under the main agent (lead) or under one added with lead: true.",
+          "delivered": {}
+        },
+        {
+          "result": "No agent called ghost in this team. The team: lead, api, routes.",
+          "delivered": {}
+        },
+        {
+          "result": "There is already an agent called parent in this team. Choose another name.",
+          "delivered": {}
+        },
+        {
+          "result": "Sent to api.",
+          "delivered": {
+            "api": [
+              "[Team message from routes to api:]\nRoutes are in."
+            ]
+          }
+        },
+        {
+          "result": "You are the main agent: you have no parent.",
+          "delivered": {}
+        },
+        {
+          "delivered": {
+            "api": [
+              "[Team report from routes to api:]\nRoutes done."
+            ]
+          }
+        },
+        {
+          "size": 3,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s2",
+              "name": "routes",
+              "parent": "api",
+              "state": "finished",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: Min above 1 opens min − 1 slots from the start of the reply; each subagent added fills one, and the note says how many the person opened.",
+      "start": {
+        "min": 3,
+        "max": 4,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "user"
+        },
+        {
+          "op": "tree"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "tree"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "web",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "docs",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "note": "<session-note>\nThe team may have 3 to 4 agents, you included.\nWork through subagents from the start: you lead, and they do the work. The person opened 2 slots for them: fill them with team_add.\n</session-note>"
+        },
+        {
+          "size": 1,
+          "open_slots": 2,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        },
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "size": 2,
+          "open_slots": 1,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        },
+        {
+          "result": "Added web (owns a part): it is working. Its report comes to you when it finishes. The team is 3 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added docs (owns a part): it is working. Its report comes to you when it finishes. The team is 4 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "size": 4,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s2",
+              "name": "web",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s3",
+              "name": "docs",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: a subagent's mode is its parent's or stricter; request options only with a stricter one; a main agent made stricter reaches down the tree.",
+      "start": {
+        "min": 1,
+        "max": 6,
+        "capacity": {},
+        "mode": "edit"
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "goal": "Ship v2",
+            "mode": "auto"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "mode": "manual",
+            "request": [
+              "write"
+            ],
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "api",
+            "mode": "edit",
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "api",
+            "mode": "plan",
+            "request": [
+              "execute"
+            ],
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "web",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "request": [
+              "write"
+            ],
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "web",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "tree"
+        },
+        {
+          "op": "main_mode",
+          "mode": "manual"
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "A subagent's mode cannot be looser than its parent's (edit).",
+          "delivered": {}
+        },
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "A subagent's mode cannot be looser than its parent's (manual).",
+          "delivered": {}
+        },
+        {
+          "result": "Added routes (owns a part): it is working. Its report comes to you when it finishes. The team is 3 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Request options go with a mode stricter than the parent's: without one there is nothing to ask for.",
+          "delivered": {}
+        },
+        {
+          "result": "Added web (owns a part): it is working. Its report comes to you when it finishes. The team is 4 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "size": 4,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "manual",
+              "request": [
+                "write"
+              ]
+            },
+            {
+              "id": "s2",
+              "name": "routes",
+              "parent": "api",
+              "state": "working",
+              "mode": "plan",
+              "request": [
+                "execute"
+              ]
+            },
+            {
+              "id": "s3",
+              "name": "web",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": []
+        },
+        null,
+        {
+          "size": 4,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "manual",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "manual",
+              "request": [
+                "write"
+              ]
+            },
+            {
+              "id": "s2",
+              "name": "routes",
+              "parent": "api",
+              "state": "working",
+              "mode": "plan",
+              "request": [
+                "execute"
+              ]
+            },
+            {
+              "id": "s3",
+              "name": "web",
+              "parent": "lead",
+              "state": "working",
+              "mode": "manual",
+              "request": []
+            }
+          ],
+          "ended": []
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: team_whoami says who the caller is, its mode and request options, its limits and its context use; the main agent has no parent or task line.",
+      "start": {
+        "min": 1,
+        "max": 4,
+        "capacity": {},
+        "mode": "edit"
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "parser",
+            "role": "reviews the parser",
+            "task": "Review src/parse.ts.\nReport what is wrong.",
+            "goal": "Ship v2",
+            "mode": "manual",
+            "request": [
+              "write"
+            ]
+          }
+        },
+        {
+          "op": "context",
+          "by": "parser",
+          "used": 1000,
+          "window": 4000
+        },
+        {
+          "op": "call",
+          "by": "parser",
+          "tool": "team_whoami",
+          "args": {}
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_whoami",
+          "args": {}
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added parser (reviews the parser): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        null,
+        {
+          "result": "You are parser (s1): reviews the parser.\nYour parent: lead.\nYour task: Review src/parse.ts.\nYour mode: manual.\nYou may ask the main agent for: write (team_request).\nYour model: m.\nLimits: the team is 2 of 4 agents; at most 50 steps a turn; only the main agent adds subagents.\nContext: 1000 of 4000 tokens (25%).",
+          "delivered": {}
+        },
+        {
+          "result": "You are lead (lead): the main agent.\nYour mode: edit.\nYour model: m.\nLimits: the team is 2 of 4 agents; at most 200 steps a turn; you add subagents.",
+          "delivered": {}
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: team_finish reports to the parent and ends the subagent: it leaves the size, its name is free again with a new id, and it stays in the tree's ended list.",
+      "start": {
+        "min": 1,
+        "max": 4,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_finish",
+          "args": {
+            "report": "All done: see src/api.ts:12."
+          }
+        },
+        {
+          "op": "tree"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_send",
+          "args": {
+            "to": [
+              "api"
+            ],
+            "message": "One more thing."
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Reported to lead. Your task is done: stop here.",
+          "delivered": {
+            "lead": [
+              "[Team report from api to lead:]\nAll done: see src/api.ts:12."
+            ]
+          }
+        },
+        {
+          "size": 1,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": [
+            {
+              "id": "s1",
+              "name": "api",
+              "state": "done"
+            }
+          ]
+        },
+        {
+          "result": "No agent called api in this team. The team: lead.",
+          "delivered": {}
+        },
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "size": 2,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s2",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": [
+            {
+              "id": "s1",
+              "name": "api",
+              "state": "done"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: findings are kept beyond an agent's end; the agent, the subagents above it and the main agent read them; nobody else.",
+      "start": {
+        "min": 1,
+        "max": 6,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "parent": "api"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "web",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "call",
+          "by": "routes",
+          "tool": "team_findings",
+          "args": {
+            "action": "write",
+            "text": "Login is POST /session."
+          }
+        },
+        {
+          "op": "call",
+          "by": "routes",
+          "tool": "team_findings",
+          "args": {
+            "action": "write",
+            "text": "Tokens expire after 1h."
+          }
+        },
+        {
+          "op": "call",
+          "by": "routes",
+          "tool": "team_findings",
+          "args": {
+            "action": "read"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_findings",
+          "args": {
+            "action": "read",
+            "of": "routes"
+          }
+        },
+        {
+          "op": "call",
+          "by": "web",
+          "tool": "team_findings",
+          "args": {
+            "action": "read",
+            "of": "routes"
+          }
+        },
+        {
+          "op": "call",
+          "by": "web",
+          "tool": "team_findings",
+          "args": {
+            "action": "read"
+          }
+        },
+        {
+          "op": "call",
+          "by": "routes",
+          "tool": "team_finish",
+          "args": {
+            "report": "Done."
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_findings",
+          "args": {
+            "action": "read",
+            "of": "s2"
+          }
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added routes (owns a part): it is working. Its report comes to you when it finishes. The team is 3 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added web (owns a part): it is working. Its report comes to you when it finishes. The team is 4 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Noted (finding 1).",
+          "delivered": {}
+        },
+        {
+          "result": "Noted (finding 2).",
+          "delivered": {}
+        },
+        {
+          "result": "Findings of routes (s2):\n1. Login is POST /session.\n2. Tokens expire after 1h.",
+          "delivered": {}
+        },
+        {
+          "result": "Findings of routes (s2):\n1. Login is POST /session.\n2. Tokens expire after 1h.",
+          "delivered": {}
+        },
+        {
+          "result": "You may read only your own findings and those of the subagents under you.",
+          "delivered": {}
+        },
+        {
+          "result": "Findings of web (s3):\n(no findings)",
+          "delivered": {}
+        },
+        {
+          "result": "Reported to api. Your task is done: stop here.",
+          "delivered": {
+            "api": [
+              "[Team report from routes to api:]\nDone."
+            ]
+          }
+        },
+        {
+          "result": "Findings of routes (s2):\n1. Login is POST /session.\n2. Tokens expire after 1h.",
+          "delivered": {}
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: when a subagent's context fills it is paused and given one final turn; a message to it then stays in the room; when that turn ends it is out of context and its parent gets the notice and its last reply as its report. The main agent replaces it: the same name and position, a new id, briefed with the task and the findings.",
+      "start": {
+        "min": 1,
+        "max": 4,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Write the API.",
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "api",
+          "tool": "team_findings",
+          "args": {
+            "action": "write",
+            "text": "Half the routes are done."
+          }
+        },
+        {
+          "op": "context_full",
+          "by": "api"
+        },
+        {
+          "op": "state"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_send",
+          "args": {
+            "to": [
+              "api"
+            ],
+            "message": "How far?"
+          }
+        },
+        {
+          "op": "finish",
+          "by": "api",
+          "reply": "Saved; the rest is the auth routes."
+        },
+        {
+          "op": "tree"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_replace",
+          "args": {
+            "name": "api"
+          }
+        },
+        {
+          "op": "started",
+          "by": "api"
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Noted (finding 1).",
+          "delivered": {}
+        },
+        {
+          "delivered": {
+            "api": [
+              "[Team notice to api: your context is full]\nSave what you found with team_findings, then tell your parent with team_send that you are gone and what is left undone. This is your last turn: at most 8 steps."
+            ]
+          }
+        },
+        [
+          {
+            "name": "lead",
+            "manager": null,
+            "state": "working"
+          },
+          {
+            "name": "api",
+            "manager": "lead",
+            "state": "paused"
+          }
+        ],
+        {
+          "result": "Sent to api.",
+          "delivered": {}
+        },
+        {
+          "delivered": {
+            "lead": [
+              "[Team report from api to lead:]\napi (s1) ran out of context and has ended. Its findings: team_findings with of api.\n\nSaved; the rest is the auth routes."
+            ]
+          }
+        },
+        {
+          "size": 1,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": [
+            {
+              "id": "s1",
+              "name": "api",
+              "state": "out_of_context"
+            }
+          ]
+        },
+        {
+          "result": "Replaced api: s2 takes its place, briefed with its task and findings. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "text": "Write the API.\n\nYou take over from api (s1), which ended. What it found:\n1. Half the routes are done."
+        },
+        {
+          "size": 2,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s2",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": [
+            {
+              "id": "s1",
+              "name": "api",
+              "state": "out_of_context"
+            }
+          ]
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: a narrowed subagent asks the main agent with team_request (only for a kind in its options); the main agent decides once, and the subagent is told under the decision mark.",
+      "start": {
+        "min": 1,
+        "max": 4,
+        "capacity": {},
+        "mode": "auto"
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "ops",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "goal": "Ship v2",
+            "mode": "edit",
+            "request": [
+              "execute"
+            ]
+          }
+        },
+        {
+          "op": "call",
+          "by": "ops",
+          "tool": "team_request",
+          "args": {
+            "tool": "edit",
+            "pattern": "src/a.ts",
+            "why": "to fix it"
+          }
+        },
+        {
+          "op": "call",
+          "by": "ops",
+          "tool": "team_request",
+          "args": {
+            "tool": "bash",
+            "pattern": "git push *",
+            "why": "to publish the branch"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_decide",
+          "args": {
+            "id": "r1",
+            "allow": true,
+            "note": "Only to origin."
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_decide",
+          "args": {
+            "id": "r1",
+            "allow": false
+          }
+        },
+        {
+          "op": "call",
+          "by": "ops",
+          "tool": "team_request",
+          "args": {
+            "tool": "bash",
+            "pattern": "rm -rf build",
+            "why": "clean"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_decide",
+          "args": {
+            "id": "r2",
+            "allow": false
+          }
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added ops (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 4 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "You may not ask for edit: your request options are execute.",
+          "delivered": {}
+        },
+        {
+          "result": "Asked the main agent (r1): you are told its decision. Carry on with what you can meanwhile.",
+          "delivered": {
+            "lead": [
+              "[Team request r1 from ops to lead:]\nbash git push * — to publish the branch"
+            ]
+          }
+        },
+        {
+          "result": "Allowed r1 for ops.",
+          "delivered": {
+            "ops": [
+              "[Team decision on r1 from lead to ops:]\nAllowed: bash git push *. Try it again now.\n\nOnly to origin."
+            ]
+          }
+        },
+        {
+          "result": "No open request r1.",
+          "delivered": {}
+        },
+        {
+          "result": "Asked the main agent (r2): you are told its decision. Carry on with what you can meanwhile.",
+          "delivered": {
+            "lead": [
+              "[Team request r2 from ops to lead:]\nbash rm -rf build — clean"
+            ]
+          }
+        },
+        {
+          "result": "Refused r2 for ops.",
+          "delivered": {
+            "ops": [
+              "[Team decision on r2 from lead to ops:]\nRefused: bash rm -rf build."
+            ]
+          }
+        }
+      ]
+    },
+    {
+      "about": "Spec 3.4.0: team_assign gives a subagent new work (it is told, and woken), may move it under another sub-manager, never under its own subtree, and never to one that ended.",
+      "start": {
+        "min": 1,
+        "max": 6,
+        "capacity": {}
+      },
+      "steps": [
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "api",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "goal": "Ship v2"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "routes",
+            "role": "owns a part",
+            "task": "Do the part.",
+            "lead": true,
+            "parent": "api"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_add",
+          "args": {
+            "name": "web",
+            "role": "owns a part",
+            "task": "Do the part."
+          }
+        },
+        {
+          "op": "finish",
+          "by": "web",
+          "reply": "Web done."
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_assign",
+          "args": {
+            "name": "web",
+            "task": "Now the docs.",
+            "role": "writes the docs"
+          }
+        },
+        {
+          "op": "state"
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_assign",
+          "args": {
+            "name": "api",
+            "task": "x",
+            "parent": "routes"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_assign",
+          "args": {
+            "name": "web",
+            "task": "x",
+            "parent": "routes",
+            "mode": "auto"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_assign",
+          "args": {
+            "name": "web",
+            "task": "Help with routes.",
+            "parent": "routes"
+          }
+        },
+        {
+          "op": "call",
+          "by": "web",
+          "tool": "team_finish",
+          "args": {
+            "report": "ok"
+          }
+        },
+        {
+          "op": "call",
+          "by": "lead",
+          "tool": "team_assign",
+          "args": {
+            "name": "web",
+            "task": "again"
+          }
+        },
+        {
+          "op": "tree"
+        }
+      ],
+      "expect": [
+        {
+          "result": "Added api (owns a part): it is working. Its report comes to you when it finishes. The team is 2 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added routes (owns a part): it is working. Its report comes to you when it finishes. The team is 3 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "result": "Added web (owns a part): it is working. Its report comes to you when it finishes. The team is 4 of 6 agents.",
+          "delivered": {}
+        },
+        {
+          "delivered": {
+            "lead": [
+              "[Team report from web to lead:]\nWeb done."
+            ]
+          }
+        },
+        {
+          "result": "Assigned to web: it is told now.",
+          "delivered": {
+            "web": [
+              "[Team assignment from lead to web:]\nNow the docs."
+            ]
+          }
+        },
+        [
+          {
+            "name": "lead",
+            "manager": null,
+            "state": "working"
+          },
+          {
+            "name": "api",
+            "manager": "lead",
+            "state": "working"
+          },
+          {
+            "name": "routes",
+            "manager": "api",
+            "state": "working"
+          },
+          {
+            "name": "web",
+            "manager": "lead",
+            "state": "working"
+          }
+        ],
+        {
+          "result": "api cannot go under routes: that is under it.",
+          "delivered": {}
+        },
+        {
+          "result": "A subagent's mode cannot be looser than its parent's (edit).",
+          "delivered": {}
+        },
+        {
+          "result": "Assigned to web: it is told now.",
+          "delivered": {
+            "web": [
+              "[Team assignment from lead to web:]\nHelp with routes."
+            ]
+          }
+        },
+        {
+          "result": "Reported to routes. Your task is done: stop here.",
+          "delivered": {
+            "routes": [
+              "[Team report from web to routes:]\nok"
+            ]
+          }
+        },
+        {
+          "result": "web has ended: replace it (team_replace) or add another.",
+          "delivered": {}
+        },
+        {
+          "size": 3,
+          "open_slots": 0,
+          "agents": [
+            {
+              "id": "lead",
+              "name": "lead",
+              "parent": null,
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s1",
+              "name": "api",
+              "parent": "lead",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            },
+            {
+              "id": "s2",
+              "name": "routes",
+              "parent": "api",
+              "state": "working",
+              "mode": "edit",
+              "request": []
+            }
+          ],
+          "ended": [
+            {
+              "id": "s3",
+              "name": "web",
+              "state": "done"
+            }
+          ]
+        }
+      ]
     }
   ]
 }
diff --git a/harness/subagents.md b/harness/subagents.md
index 59640a8..3245907 100644
--- a/harness/subagents.md
+++ b/harness/subagents.md
@@ -266,6 +266,9 @@ a message wakes it — 3.2.0), **`paused`** (its context is full; its last turn
 **An ended subagent** — `done`, `out_of_context`, `stopped` — leaves the size and its slot frees;
 its name may be used again; it keeps its id, and the tree keeps it (§12) with its findings and
 transcript for the rest of the session. Messages to it by name are refused as `unknown`.
+The subagents under a sub-manager that ended by itself (`done`, `out_of_context`) move under its
+parent until the main agent replaces it, and then back under the replacement; a stopped one's
+subtree is stopped with it (§10).
 
 **When its context fills.** A subagent does not compact. When its context reaches
 `team.context.pause_at` of its model's window (default 0.85, the fraction compaction would start
@@ -348,3 +351,12 @@ the link (`acp.md`, capability `tree`):
 - A subagent: `blocks/team_member.md` (3.4.0: its id and position, `team_whoami`, `team_findings`,
   `team_finish`, that its context is limited and it will be told when it fills); a sub-manager,
   `blocks/team_submanager.md` in place of the whole lead block it got in 3.2.0.
+
+## 14. In an agent chat
+
+The main agent of an agent chat is the CLI on the linked device: it starts and runs the
+subagents there, under every rule above. The server carries the chat's Subagents setting down with
+every turn and the tree, approvals, findings and transcripts up, and offers no subagents of its own
+to an agent chat. The protocol: `acp.md`, *Subagents in an agent chat* (capabilities `agents` and
+`tree`).
+