pi-subagentura

Note: The docs/ folder is managed by the pi-docs package.

A public Pi package that combines lightweight in-process delegation with observable, attachable interactive sub-agents.

Interactive sub-agents are the standout feature. subagent_interactive launches a real, separate Pi session in a tmux or zellij pane. Watch its Pi TUI and tool activity in real time, focus the pane from your current multiplexer, attach from another terminal, and send follow-up prompts into the same child context. Every turn also writes durable lifecycle and output artifacts, so the work remains inspectable after detach, reload, or parent restart.

Available tools:

• subagent_with_context — spawn a sub-agent that inherits the full conversation history
• subagent_isolated — spawn a sub-agent with a fresh, empty context window
• get_subagent_status — poll an async subagent job for live progress
• get_subagent_result — block until an async job completes and return the final output
• cancel_subagent — abort a running async job
• prune_subagent_jobs — remove all completed and failed jobs from the registry
• subagent_interactive — spawn an attachable Pi session in a tmux/zellij pane
• get_interactive_subagent_status — list attachable sessions with pane/session/artifact metadata
• cancel_interactive_subagent — kill an attachable interactive sub-agent pane
• send_interactive_subagent_message — send a follow-up into a live child REPL without losing context
• read_subagent_artifact — read an interactive sub-agent's lifecycle events and output
• list_subagent_artifacts — list known interactive sub-agents and artifact locations

For quick delegated work, in-process sub-agents stream progress in the current Pi process. With async: true, they run in the background while the main agent continues, and can be polled, collected, or cancelled by job id.

Interactive children remain idle between turns instead of exiting. Follow-ups preserve model context, and each completion maps its Pi turnId to an immutable outputs/<eventId>.md snapshot recoverable through read_subagent_artifact.

Why use it?

• Watch a real child Pi session and its tool activity live in tmux or zellij
• Focus a pane locally or attach from another terminal at any time
• Continue true follow-up turns without losing the child's model context
• Inspect durable per-turn outputs and lifecycle events after detach or restart
• Run lightweight sub-agents in-process or in the background
• Compare context-aware and isolated reasoning
• Poll, collect, or cancel background jobs on demand

Installation

Install globally:

pi install npm:pi-subagentura

Install for just the current project:

pi install -l npm:pi-subagentura

Try it for a single run without installing:

pi -e npm:pi-subagentura

You can also install directly from GitHub:

pi install git:github.com/lmn451/pi-subagentura

Tools

subagent_with_context

Starts a sub-agent with the current conversation history included in its prompt.

Parameters:

• task — required task for the sub-agent
• persona — optional system-style persona
• model — optional model override like anthropic/claude-sonnet-4-5
• cwd — optional working directory override
• async — run in background; returns a jobId immediately instead of blocking
• notifyOnComplete — "inject" (default for async) persists full output; "notify" persists a pointer only. Both modes show a user-facing completion notification.
• triggerTurnOnComplete — optional override; notify defaults false and inject defaults true
• maxAge — optional TTL in ms for completed job retention (async only)

Best for:

• review tasks that depend on prior discussion
• continuing a line of reasoning in parallel
• focused implementation or research using the current context
• background side-quests that report results later

subagent_isolated

Starts a sub-agent with no inherited conversation history.

Parameters:

• task — required task for the sub-agent
• persona — optional system-style persona
• model — optional model override like anthropic/claude-sonnet-4-5
• cwd — optional working directory override
• async — run in background; returns a jobId immediately instead of blocking
• notifyOnComplete — "inject" (default for async) persists full output; "notify" persists a pointer only. Both modes show a user-facing completion notification.
• triggerTurnOnComplete — optional override; notify defaults false and inject defaults true
• maxAge — optional TTL in ms for completed job retention (async only)

Async spawn results state whether completion output will be injected into the parent LLM and whether delivery will automatically start a new parent turn.

Best for:

• second opinions
• clean-room summaries
• avoiding context contamination from the parent session
• background analysis without polluting the main conversation

Async Workflow Tools

When you spawn a sub-agent with async: true, it returns a jobId immediately and runs in the background. Use these tools to manage async jobs:

get_subagent_status

Poll an async subagent job by jobId. Returns a live preview of the subagent's current turn, active tool, and partial output.

Parameters:

• jobId — required job ID returned by the async spawn

get_subagent_result

Block until an async subagent job completes, then return the final output and usage summary. If the job is already done, it returns immediately.

Parameters:

• jobId — required job ID returned by the async spawn

cancel_subagent

Abort a running async subagent job by jobId.

Parameters:

• jobId — required job ID returned by the async spawn

prune_subagent_jobs

Remove all completed and failed subagent jobs from the registry. Running and cancelled jobs are preserved.

Interactive Sub-agent Tools

Observability and attachability are the primary design goals of interactive sub-agents—not debugging afterthoughts. Each child is a separate Pi process in a tmux or zellij pane: watch it live, focus it from the current mux, attach from another terminal, or send follow-ups through the parent while preserving child context. If the parent is outside a mux, the child starts in a detached session and returns an attach command. The pane is the live view; durable artifacts are the source of truth.

subagent_interactive

Starts a separate interactive pi process in a tmux/zellij pane and returns immediately with:

• sub-agent id
• pane id and mux backend (tmux or zellij)
• attach command (works from outside the mux session)
• focus command (works from inside the same mux session)
• child Pi session file path
• artifact directory (events.ndjson + output.md)
• the window/tab name (in background mode) so you can find it in your mux UI

Parameters:

• task — required initial task
• name — optional display name for the pane/session
• persona — optional system prompt appended to the child session
• model — optional model override
• cwd — optional working directory
• includeContext — include serialized parent conversation in the child prompt (default: false)
• mux — optional backend: "auto" (default), "tmux", or "zellij". Auto picks the currently attached mux (via ZELLIJ_SESSION_NAME / TMUX env vars) then falls back to whichever backend binary is available. Explicit choice forces that backend.
• background — spawn in a detached named window/tab (invisible) instead of a visible horizontal split. Default true — your mux layout is undisturbed and you can attach later with the returned focus command. Pass background: false for a side-by-side split you can watch in real time.
• notifyOnComplete — "inject" (default) persists full output; "notify" persists only an artifact pointer. Both modes show a user-facing completion notification.
• triggerTurnOnComplete — optional override. Notify defaults to false; inject defaults to true.

The spawn result states whether completion output will be injected into the parent LLM and whether delivery will automatically start a new parent turn.

The sub-agent's work is always written to the artifact dir as events.ndjson (lifecycle log) and output.md (clean prose the child writes). The pane is for live monitoring; the artifact is the source of truth. The artifact survives parent restarts — sub-agents that finish while you're away are picked up on the next poll.

The interactive sub-agent registry state survives parent reloads and restarts. When spawned, a per-(cwd) state file is written to <cwd>/.pi/subagentura-state.json.

The state file and subagent panes are preserved across these actions:

| Action | State file | Panes | Rehydrated next start? | | ------------------------------------------------- | ----------- | ---------- | ---------------------------------------- | | Ctrl+D (quit) → restart with --session/-r | Kept | Preserved | ✅ Same session, parentSessionId matches | | Ctrl+D → fresh pi (no session) | Kept | Preserved | ❌ Different session, no match | | /reload | Kept | Preserved | ✅ Same session | | /resume (switch to another session) | Kept | Preserved | ✅ If parentSessionId matches | | /new | Deleted | Killed | ❌ Clean slate | | /fork | Deleted | Killed | ❌ Clean slate |

Note: /new deletes the state file. If you do /new and then /resume back to the session where subagents were spawned, they will not reappear — the state file was already deleted. Only /reload or a restart with the same session (--session/-r) preserves the registry.

On /reload and /resume, the session_start handler rehydrates the in-memory registry, filtering by parentSessionId so only subagents that were created in the current session are restored. Protocol-v2 byte cursors, pending delivery intents, and receipts are restored so reload does not unconditionally replay already-dispatched completions.

Implementation details for crash-safe ordering and delivery recovery are in the state-file invariants in the source repository.

Sub-agent completion protocol

Every interactive child runs protocol-only Pi lifecycle hooks and therefore requires Pi SDK >=0.80.6 <0.81.0. before_agent_start creates a provisional turn, the first turn_start binds it to the persisted Pi user-entry id, tool hooks record activity, and agent_settled records the authoritative completion after retries, compaction, and queued continuations. When Pi accepts Enter while streaming, it treats the message as steering inside the current agent run and does not emit another before_agent_start. The child protocol therefore detects the newly persisted steering user entry before its provider request and starts a distinct artifact turn for it. The explicit CLI remains supported:

$ARTIFACT_DIR/cli.mjs done 0       # success — parent reads the literal output.md path baked into the child prompt
$ARTIFACT_DIR/cli.mjs error "msg"  # unrecoverable failure
# 'cancelled' is only set by the parent via cancel_interactive_subagent

The explicit completion command is mandatory for every initial and follow-up turn. The child must complete these steps in order: write the final result to output.md, run cli.mjs done 0, wait until exactly one completion event is recorded successfully, then send its final assistant response. The command must be the final tool call of the turn. If it fails to execute, the child must not finalize; it fixes the failure and retries until completion is recorded. The child lifecycle hook at agent_settled is a crash-safety fallback, not a substitute for the explicit command. The system prompt, initial task footer, and every injected follow-up prompt repeat this requirement so the command remains the model's most recent instruction.

At each child turn start, mutable output.md is atomically reset without touching earlier snapshots. Before each completion event, the current staging file (including an empty file when the turn wrote nothing) is copied atomically to outputs/<eventId>.md with byte count and SHA-256 metadata. Events are consumed in physical NDJSON byte order; timestamps are display-only. Mixed v1/v2 logs and legacy output-N.md snapshots remain readable. New legacy completions are pointer-only because mutable output.md cannot be safely attributed to a turn.

Immutable snapshots are limited to 1 MiB. The parent and generated child CLI check the staging file size before reading it. If output.md exceeds the limit, the completion is still recorded with outputError.code = "output_too_large" and its observed byte count, but no immutable snapshot is created or injected into parent context. The oversized staging file remains available for manual inspection in the artifact directory.

Completion notifications: notify vs inject

Completion delivery has two independent settings. notifyOnComplete controls the payload saved for parent LLM context (full output versus artifact pointer); triggerTurnOnComplete independently controls whether delivery starts a parent LLM turn. Both payload modes display the same user-facing notification; notifyOnComplete does not control toast behavior.

• "inject" (default, including async in-process tools) persists one attributed custom completion message and triggers a turn by default. Explicit triggerTurnOnComplete: false disables triggering.
• "notify" persists an attributed pointer-only custom message in parent context without triggering a provider call. Explicit triggerTurnOnComplete: true wakes the parent immediately.

| Configuration | Persisted in parent context | Starts an immediate parent turn | | ----------------------------------------- | --------------------------- | ------------------------------- | | inject (default) | Full completion output | Yes | | inject + triggerTurnOnComplete: false | Full completion output | No | | notify (default trigger behavior) | Artifact pointer only | No | | notify + triggerTurnOnComplete: true | Artifact pointer only | Yes |

Therefore plain notify records the completion for both the UI and the parent conversation, but the LLM does not react immediately. The pointer becomes available to the model when the user starts the next parent turn. It is not a visual-only notification and the completion is not discarded.

Both modes also show a user notification (ui.notify) after successful delivery. The notification explicitly states whether completion output was injected into the parent LLM and whether a new parent turn will start automatically. Async in-process and interactive spawn results state the same behavior in future tense before work begins. Errors use an error notification, cancellations use a warning, and successful completions use an informational notification.

In the TUI, a normal completion therefore produces two visually separate elements (exact colors depend on the active theme):

1. A user-notification status line (ui.notify). This is UI-only and is never added to session or LLM context.
2. A custom-message block beginning with [Sub-agent ...]. This is the separately persisted LLM-facing delivery: full output in inject mode, or only status and artifact pointers in notify mode.

Parent-initiated cancellation is the exception: the cancel tool result already acknowledges it to the LLM, so cancellation shows the UI notification without adding a duplicate custom-message block.

LLM-facing completions wait while the parent streams and flush as one bounded message after agent_settled. The durable FIFO is limited to 32 records / 256 KiB, with 32 KiB output per record and 64 KiB per flush. Overflow keeps status and artifact pointers rather than silently dropping completion. Delivery is at-least-once: deterministic delivery IDs and session receipts prevent routine reload replay, but a crash between synchronous dispatch and receipt persistence can still duplicate one completion.

get_interactive_subagent_status

Lists tracked interactive sub-agents, attach/select commands, and session paths. It intentionally does not capture pane output to avoid consuming model context.

Parameters:

• jobId — optional interactive sub-agent id; omit it to list all tracked sessions

cancel_interactive_subagent

Kills the mux pane for an interactive sub-agent by id. Writes a cancelled event and immutable output snapshot before killing the pane, so artifacts remain self-describing. Because the cancel tool result already acknowledges the action to the parent LLM, parent-initiated cancellation also shows a warning via ui.notify and persists a delivery receipt without injecting a duplicate cancellation completion. The receipt is written before the pane is killed, so a later poll or restart cannot replay that cancellation into LLM context. Child-originated or otherwise unacknowledged cancellations still use the normal completion-delivery path.

Parameters:

• jobId — required interactive sub-agent id returned by subagent_interactive

send_interactive_subagent_message

Sends a follow-up prompt to a running or idle interactive sub-agent by id. The message is delivered into the child's existing REPL via the sub-agent's mux backend (tmux send-keys or zellij write-chars + write 13), so the child's model context is preserved — this is a true follow-up turn, not a fresh spawn. A message submitted while the child streams is processed through Pi's one-at-a-time steering queue; its persisted user entry receives a distinct artifact turn and immutable completion snapshot. The child will call cli.mjs done 0 again when finished, which wakes the parent through the usual notifyOnComplete path. Refuses to send if the sub-agent is not in the registry, is neither running nor idle, or if the mux itself rejects the send call (e.g. the pane was killed between the status check and send). All three failure modes return a structured isError: true result.

Parameters:

• id — required sub-agent id
• message — required follow-up prompt text

list_subagent_artifacts

Lists all known interactive sub-agents: id, name, status, artifact directory, and last-update timestamp. Use this to discover sub-agents that finished while the parent was away.

read_subagent_artifact

Reads a sub-agent's artifact by id. Returns the lifecycle event log (pass since to fetch only new events) and, by default, the sub-agent's output.md content (the latest turn's output). This is the canonical way to get the sub-agent's work product — the parent agent does not need to read the tmux pane or capture rendered TUI.

Protocol-v2 completions map each Pi-derived turnId to an immutable outputs/<eventId>.md snapshot. Pass turnId to read that output; the response's details.outputHistory lists the available turnId/eventId mappings. Legacy output-N.md history remains available through numeric turn and details.availableTurns.

Parameters:

• id — required sub-agent id
• since — optional unix-ms timestamp; only return events with ts >= since
• includeOutput — include the output (default true); historical selectors imply output
• turn — optional turn number; read output-N.md for that specific turn instead of the latest output.md
• turnId — optional protocol-v2 Pi turn id; read its immutable outputs/<eventId>.md snapshot

list_available_models

List all available AI models with auth status. Use this to validate model identifiers before passing them to subagent tools — prevents silent fallback to the parent session model.

Parameters:

• filter — optional substring filter for provider or model name
• authOnly — if true (default), only return models with configured auth

Example prompts

• “Use a sub-agent to review this change and list risks.”
• “Use an isolated sub-agent to propose a README outline for this repo.”
• “Spawn a context-aware sub-agent to continue debugging while we keep planning here.”
• “Run a sub-agent in the background to run the test suite, then notify me when done.”
• “Spawn two isolated async sub-agents to review this code from different angles, then collect both results.”
• “Start an interactive sub-agent in tmux for investigating the auth bug; give me the attach command.”
• “Open an interactive sub-agent in a visible zellij pane so I can watch its tool calls live.”
• “Attach to the existing interactive sub-agent and send it a follow-up without losing context.”

Development

This repo uses npm for local development.

npm install
npm test
npm run pack:check

Debug logging

Set SUBAGENT_DEBUG_LOG_DIR=/some/path to write a JSONL trace of sub-agent lifecycle events to debug-YYYY-MM-DD.jsonl in that directory. Each line is a self-describing JSON object with timestamp, level, event, and event-specific fields.

The tool_start event records the toolName and full args of every tool the sub-agent invokes — useful for replaying or auditing what a sub-agent did. Other events cover session creation, turns, message updates, prompts, and job completion.

The feature is a no-op when the env var is unset.

SUBAGENT_DEBUG_LOG_DIR=./.pi-debug pi   # writes ./pi-debug/debug-2026-06-10.jsonl

Contributing

Contributions are welcome. See CONTRIBUTING.md.

A pre-commit hook formats staged files (via simple-git-hooks + lint-staged). A pre-push hook runs npm run format:check across the repository.

Install or refresh the hooks with npm run hooks:install. To skip a hook once, set SKIP_SIMPLE_GIT_HOOKS=1. To reformat the repository:

npm run format.

License

MIT