@deepseek-ai/dsh-tool-subagent-control

English | 中文

The optional, globally named send_message, interrupt_agent, and list_agents tools are thin adapters over ctx.subagents. Provider-bound @deepseek-ai/dsh-tool-subagent instances register distinct delegation tools per transport; this separately loaded package registers shared control tools once, so multiple delegation tools never register duplicate global controls. The root plugin registers send_message and interrupt_agent and requires only subagents; the separately loadable ./list-agents plugin registers list_agents and declares subagents plus agents as load-time dependencies. Its catalog reads additionally require the session store and projection registry at call time, but no query service. A deployment can keep the root tools while omitting the list tool. No tool's presence determines whether a delegation tool starts continuable work. These tools own only the parent-to-child direction; the independently installed @deepseek-ai/dsh-tool-subagent-report owns the child-to-parent direction.

The tool performs no lifecycle routing — residency and cold resume belong to the subagent service. It passes exec.agent as the exact live parent that authorizes delivery and records every message source as { kind: 'coordinator', senderSessionId: parent.id }, which the service retains but never treats as authority. Every message becomes the subagent's next FIFO turn through Agent.followup(): if the child is still working, the message waits until its current turn finishes, so it cannot redirect work already underway. The tool forwards its execution signal, which owns admission only until inbox acceptance; once the child accepts the message the accepted turn cannot be cancelled through this tool. This call returns no child reply — its transcript by that id is the source of what it did — and a child with report sends content on its own initiative as a separate parent message. A delivery failure becomes an errored tool result stating the message was not delivered.

interrupt_agent(agent_id) passes exec.agent as the exact live ancestor authority for ctx.subagents.interrupt(): the target may be a direct child or a deeper descendant, and the service — never this tool — verifies the caller against the target Activation's recorded lineage. Only the target's current turn stops (keepInbox): queued messages stay parked until a later send_message, published descendants keep running, and the child stays available for follow-ups. The call returns as soon as the stop request is accepted, without waiting for target quiescence; an absent or already-settled target is an accepted no-op, while self, sibling, stale, and non-ancestor callers become errored results.

list_agents takes one optional scope argument, derives the root id from the calling agent, and projects the service catalog to continuable children without a cursor. The default children scope reads ctx.subagents.listChildren(); descendants reads ctx.subagents.listDescendants(), whose one-corpus walk crosses ordinary sessions and one-shot children and renders surviving rows in stable pre-order with parent=<id> depth=<n>. The parent annotation is the durable direct-parent session id and may name an ordinary session omitted from the output. For the calling agent, only depth-1 child entries are send_message candidates; deeper child entries are interrupt_agent candidates only. Status comes from the live Agent registry: running (active driver), idle (resident between turns, possibly waiting on agents it started), or ready (storage only and resumable rather than terminal). The service result also contains one-shot session-backed subagents for consumers such as a UI, but those entries are omitted from this model tool because they cannot accept send_message. Diagnostics remain visible, with positions in the descendants scope. Durable identity and mode come from each child's descriptor, while delivery-time authority and Activation ownership checks remain the service's.

Model Experience

Tool schema

What the model sees

The generated schemas: send_message takes subagent_id and message, describing that the message becomes the subagent's next turn, that this call returns no answer from the subagent, and that a failure means the message was not delivered; interrupt_agent takes agent_id, describing that only the current turn stops, queued messages park, descendants keep running, and acceptance precedes the actual stop; list_agents takes the optional scope enum.

Token effect

Fixed schema cost per parent request.

KV Cache effect

Prefix-stable; the schema does not change at runtime.

Interrupt result

What the model sees

interrupt requested for agent <agent_id> on acceptance. An unauthorized caller — self, sibling, stale, or non-ancestor — is an errored result naming the rejection; an absent or settled target still renders the acceptance line.

Token effect

One short acknowledgement per call; the interrupted turn's abort is visible only in the child's own transcript.

KV Cache effect

Append-only; each result follows the reusable request prefix.

Delivery result

What the model sees

message queued as the next turn for subagent <subagent_id> on acceptance; the canonical output carries the accepted messageId. A failure — an unauthorized or unknown child, a descriptor-less child that cannot be resumed, or admission rejected — is an errored result whose message states the message was not delivered.

Token effect

One short acknowledgement per call; the child's response never returns through this call. A separately granted report may append selected content to parent history.

KV Cache effect

Append-only; newly visible content follows the reusable request prefix and does not invalidate existing KV-cache entries.

Listing result

What the model sees

One line per continuable child in stable catalog order: <id> [<status>] — <label> (running = active driver, idle = resident between turns, ready = storage only; resumable rather than terminal, not a result waiting to be collected — a direct child in that state can be resumed by send_message), plus <id> [diagnostic: <reason>] for a candidate that could not be read (corrupt, unsupported, or unavailable). The descendants scope inserts parent=<id> depth=<n> before the label dash on every line, in pre-order. One-shot children are intentionally absent; (no subagents) means no continuable child or diagnostic survived the projection. Diagnostics never expose descriptor contents.

Token effect

Grows linearly with the listed continuable children — the whole tree under the descendants scope; there is no cursor or cap, so long-lived parents with many persisted children pay the full list each call.

KV Cache effect

Append-only; each result follows the reusable request prefix.

Known Limitations and Deferred Work

• A queued message has no independent result — acceptance returns only its inbox messageId; the child's work lands in the durable child Session and is never collected through this tool. A child granted report may send selected content back separately, but that message is not this call's result.
• No steering of the current turn — every message opens a later FIFO turn, so a message sent while the child is working runs only after its current turn finishes and cannot redirect it.
• Listing is a snapshot, not a delivery promise — it may race publication, disposal, or a later message, and another process may activate a child this process reports as ready; cross-process accuracy requires a shared lease. interrupt_agent performs the authoritative live-lineage check itself, so discovery staleness cannot grant authority.
• No pagination or deletion — the complete stably ordered set is returned, and persisted children remain listed for as long as their sessions remain in persistence; a service-level bound or delete operation is a later product decision.