Sami/hermes-agent
1
0
Fork 0
forked from Zakaria/hermes-agent
Code Pull Requests Activity

Hermes-agent

Browse Source
This commit is contained in:
Zakaria
2026-06-14 14:30:48 -04:00
commit dac4b88b94
5058 changed files with 1884848 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
website/docs/user-guide/features/delegation.md
+290
View File
@@ -0,0 +1,290 @@
---
sidebar_position: 7
title: "Subagent Delegation"
description: "Spawn isolated child agents for parallel workstreams with delegate_task"
---
# Subagent Delegation
The `delegate_task` tool spawns child AIAgent instances with isolated context, restricted toolsets, and their own terminal sessions. Each child gets a fresh conversation and works independently — only its final summary enters the parent's context.
## Single Task
```python
delegate_task(
goal="Debug why tests fail",
context="Error: assertion in test_foo.py line 42",
toolsets=["terminal", "file"]
)
```
## Parallel Batch
Up to 3 concurrent subagents by default (configurable, no hard ceiling):
```python
delegate_task(tasks=[
{"goal": "Research topic A", "toolsets": ["web"]},
{"goal": "Research topic B", "toolsets": ["web"]},
{"goal": "Fix the build", "toolsets": ["terminal", "file"]}
])
```
## How Subagent Context Works
:::warning Critical: Subagents Know Nothing
Subagents start with a **completely fresh conversation**. They have zero knowledge of the parent's conversation history, prior tool calls, or anything discussed before delegation. The subagent's only context comes from the `goal` and `context` fields the parent agent populates when it calls `delegate_task`.
:::
This means the parent agent must pass **everything** the subagent needs in the call:
```python
# BAD - subagent has no idea what "the error" is
delegate_task(goal="Fix the error")
# GOOD - subagent has all context it needs
delegate_task(
goal="Fix the TypeError in api/handlers.py",
context="""The file api/handlers.py has a TypeError on line 47:
'NoneType' object has no attribute 'get'.
The function process_request() receives a dict from parse_body(),
but parse_body() returns None when Content-Type is missing.
The project is at /home/user/myproject and uses Python 3.11."""
)
```
The subagent receives a focused system prompt built from your goal and context, instructing it to complete the task and provide a structured summary of what it did, what it found, any files modified, and any issues encountered.
## Practical Examples
### Parallel Research
Research multiple topics simultaneously and collect summaries:
```python
delegate_task(tasks=[
{
"goal": "Research the current state of WebAssembly in 2025",
"context": "Focus on: browser support, non-browser runtimes, language support",
"toolsets": ["web"]
},
{
"goal": "Research the current state of RISC-V adoption in 2025",
"context": "Focus on: server chips, embedded systems, software ecosystem",
"toolsets": ["web"]
},
{
"goal": "Research quantum computing progress in 2025",
"context": "Focus on: error correction breakthroughs, practical applications, key players",
"toolsets": ["web"]
}
])
```
### Code Review + Fix
Delegate a review-and-fix workflow to a fresh context:
```python
delegate_task(
goal="Review the authentication module for security issues and fix any found",
context="""Project at /home/user/webapp.
Auth module files: src/auth/login.py, src/auth/jwt.py, src/auth/middleware.py.
The project uses Flask, PyJWT, and bcrypt.
Focus on: SQL injection, JWT validation, password handling, session management.
Fix any issues found and run the test suite (pytest tests/auth/).""",
toolsets=["terminal", "file"]
)
```
### Multi-File Refactoring
Delegate a large refactoring task that would flood the parent's context:
```python
delegate_task(
goal="Refactor all Python files in src/ to replace print() with proper logging",
context="""Project at /home/user/myproject.
Use the 'logging' module with logger = logging.getLogger(__name__).
Replace print() calls with appropriate log levels:
- print(f"Error: ...") -> logger.error(...)
- print(f"Warning: ...") -> logger.warning(...)
- print(f"Debug: ...") -> logger.debug(...)
- Other prints -> logger.info(...)
Don't change print() in test files or CLI output.
Run pytest after to verify nothing broke.""",
toolsets=["terminal", "file"]
)
```
## Batch Mode Details
When you provide a `tasks` array, subagents run in **parallel** using a thread pool:
- **Maximum concurrency:** 3 tasks by default (configurable via `delegation.max_concurrent_children` or the `DELEGATION_MAX_CONCURRENT_CHILDREN` env var; floor of 1, no hard ceiling). Batches larger than the limit return a tool error rather than being silently truncated.
- **Thread pool:** Uses `ThreadPoolExecutor` with the configured concurrency limit as max workers
- **Progress display:** In CLI mode, a tree-view shows tool calls from each subagent in real-time with per-task completion lines. In gateway mode, progress is batched and relayed to the parent's progress callback
- **Result ordering:** Results are sorted by task index to match input order regardless of completion order
- **Interrupt propagation:** Interrupting the parent (e.g., sending a new message) interrupts all active children
Single-task delegation runs directly without thread pool overhead.
## Model Override
You can configure a different model for subagents via `config.yaml` — useful for delegating simple tasks to cheaper/faster models:
```yaml
# In ~/.hermes/config.yaml
delegation:
model: "google/gemini-flash-2.0" # Cheaper model for subagents
provider: "openrouter" # Optional: route subagents to a different provider
```
If omitted, subagents use the same model as the parent.
## Toolset Selection Tips
The `toolsets` parameter controls what tools the subagent has access to. Choose based on the task:
| Toolset Pattern | Use Case |
|----------------|----------|
| `["terminal", "file"]` | Code work, debugging, file editing, builds |
| `["web"]` | Research, fact-checking, documentation lookup |
| `["terminal", "file", "web"]` | Full-stack tasks (default) |
| `["file"]` | Read-only analysis, code review without execution |
| `["terminal"]` | System administration, process management |
Certain toolsets are blocked for subagents regardless of what you specify:
- `delegation` — blocked for leaf subagents (the default). Retained for `role="orchestrator"` children, bounded by `max_spawn_depth` — see [Depth Limit and Nested Orchestration](#depth-limit-and-nested-orchestration) below.
- `clarify` — subagents cannot interact with the user
- `memory` — no writes to shared persistent memory
- `code_execution` — children should reason step-by-step
- `send_message` — no cross-platform side effects (e.g., sending Telegram messages)
## Max Iterations
Each subagent has an iteration limit (default: 50) that controls how many tool-calling turns it can take:
```python
delegate_task(
goal="Quick file check",
context="Check if /etc/nginx/nginx.conf exists and print its first 10 lines",
max_iterations=10 # Simple task, don't need many turns
)
```
## Child Timeout
By default there is **no wall-clock timeout** on subagents. Children fail only from what they're actually doing — API errors, tool errors, or hitting their iteration budget — never from a delegation-level stopwatch. Earlier releases shipped a hard cap (300s, later 600s), which kept killing legitimately busy children mid-task: deep code reviews, large research fan-outs, and slow reasoning models routinely need more than 10 minutes while making steady progress the whole time.
Genuinely stuck children are still detected: the heartbeat staleness monitor stops refreshing the parent's activity when a child makes no progress (no API calls, no tool starts), letting the gateway inactivity timeout fire on a truly wedged worker.
If you want a hard cap anyway (e.g. cost control on unattended cron-driven delegation), opt in per-install:
```yaml
delegation:
child_timeout_seconds: 0 # default: 0 = no timeout
# child_timeout_seconds: 1800 # opt-in hard cap (floor 30s)
```
A positive value enforces a hard wall-clock limit on each child; `0` or a negative value disables it.
:::tip Diagnostic dump on zero-call timeout
With a hard cap configured, if a subagent times out having made **zero** API calls (usually: provider unreachable, auth failure, or tool-schema rejection), `delegate_task` writes a structured diagnostic to `~/.hermes/logs/subagent-timeout-<session>-<timestamp>.log` containing the subagent's config snapshot, credential-resolution trace, and any early error messages. Much easier to root-cause than the previous silent-timeout behavior.
:::
## Monitoring Running Subagents (`/agents`)
The TUI ships a `/agents` overlay (alias `/tasks`) that turns recursive `delegate_task` fan-out into a first-class audit surface:
- Live tree view of running and recently-finished subagents, grouped by parent
- Per-branch cost, token, and file-touched rollups
- Kill and pause controls — cancel a specific subagent mid-flight without interrupting its siblings
- Post-hoc review: step through each subagent's turn-by-turn history even after they've returned to the parent
The classic CLI just prints `/agents` as a text summary; the TUI is where the overlay shines. See [TUI — Slash commands](/user-guide/tui#slash-commands).
## Depth Limit and Nested Orchestration
By default, delegation is **flat**: a parent (depth 0) spawns children (depth 1), and those children cannot delegate further. This prevents runaway recursive delegation.
For multi-stage workflows (research → synthesis, or parallel orchestration over sub-problems), a parent can spawn **orchestrator** children that *can* delegate their own workers:
```python
delegate_task(
goal="Survey three code review approaches and recommend one",
role="orchestrator", # Allows this child to spawn its own workers
context="...",
)
```
- `role="leaf"` (default): child cannot delegate further — identical to the flat-delegation behavior.
- `role="orchestrator"`: child retains the `delegation` toolset. Gated by `delegation.max_spawn_depth` (default **1** = flat, so `role="orchestrator"` is a no-op at defaults). Raise `max_spawn_depth` to 2 to allow orchestrator children to spawn leaf grandchildren; 3+ for deeper trees. There is no upper ceiling — cost is the practical limit.
- `delegation.orchestrator_enabled: false`: global kill switch that forces every child to `leaf` regardless of the `role` parameter.
**Cost warning:** With `max_spawn_depth: 3` and `max_concurrent_children: 3`, the tree can reach 3×3×3 = 27 concurrent leaf agents. Each extra level multiplies spend — raise `max_spawn_depth` intentionally.
## Lifetime and Durability
:::warning delegate_task is synchronous — not durable
`delegate_task` runs **inside the parent's current turn**. It blocks the parent until every child finishes (or is cancelled). It is **not** a background job queue:
- If the parent is interrupted (user sends a new message, `/stop`, `/new`), all active children are cancelled and return `status="interrupted"`. Their in-progress work is discarded.
- Children do **not** continue running after the parent turn ends.
- Cancelled children return a structured result (`status="interrupted"`, `exit_reason="interrupted"`), but because the parent was interrupted too, that result often never makes it into a user-visible reply.
For **durable long-running work** that must survive interrupts or outlive the current turn, use:
- `cronjob` (action=`create`) — schedules a separate agent run; immune to parent-turn interrupts.
- `terminal(background=True, notify_on_complete=True)` — long-running shell commands that keep running while the agent does other things.
:::
## Key Properties
- Each subagent gets its **own terminal session** (separate from the parent)
- **Nested delegation is opt-in** — only `role="orchestrator"` children can delegate further, and only when `max_spawn_depth` is raised from its default of 1 (flat). Disable globally with `orchestrator_enabled: false`.
- Leaf subagents **cannot** call: `delegate_task`, `clarify`, `memory`, `send_message`, `execute_code`. Orchestrator subagents retain `delegate_task` but still cannot use the other four.
- **Interrupt propagation** — interrupting the parent interrupts all active children (including grandchildren under orchestrators)
- Only the final summary enters the parent's context, keeping token usage efficient
- Subagents inherit the parent's **API key, provider configuration, and credential pool** (enabling key rotation on rate limits)
## Delegation vs execute_code
| Factor | delegate_task | execute_code |
|--------|--------------|-------------|
| **Reasoning** | Full LLM reasoning loop | Just Python code execution |
| **Context** | Fresh isolated conversation | No conversation, just script |
| **Tool access** | All non-blocked tools with reasoning | 7 tools via RPC, no reasoning |
| **Parallelism** | 3 concurrent subagents by default (configurable) | Single script |
| **Best for** | Complex tasks needing judgment | Mechanical multi-step pipelines |
| **Token cost** | Higher (full LLM loop) | Lower (only stdout returned) |
| **User interaction** | None (subagents can't clarify) | None |
**Rule of thumb:** Use `delegate_task` when the subtask requires reasoning, judgment, or multi-step problem solving. Use `execute_code` when you need mechanical data processing or scripted workflows.
## Configuration
```yaml
# In ~/.hermes/config.yaml
delegation:
max_iterations: 50 # Max turns per child (default: 50)
# max_concurrent_children: 3 # Parallel children per batch (default: 3)
# max_spawn_depth: 1 # Tree depth (floor 1, no ceiling, default 1 = flat). Raise to 2 to allow orchestrator children to spawn leaves; 3+ for deeper trees.
# orchestrator_enabled: true # Disable to force all children to leaf role.
model: "google/gemini-3-flash-preview" # Optional provider/model override
provider: "openrouter" # Optional built-in provider
api_mode: anthropic_messages # optional; auto-detected from base_url for anthropic_messages endpoints
# Or use a direct custom endpoint instead of provider:
delegation:
model: "qwen2.5-coder"
base_url: "http://localhost:1234/v1"
api_key: "local-key"
# api_mode: "anthropic_messages" # Optional. Wire protocol override for base_url ("chat_completions", "codex_responses", or "anthropic_messages"). Empty = auto-detect from URL (e.g. /anthropic suffix). Set explicitly for endpoints the heuristic can't classify (Azure AI Foundry, MiniMax, Zhipu GLM, LiteLLM proxies, …).
```
When `base_url` points at an Anthropic-compatible endpoint — for example a path ending in `/anthropic`, an Azure Foundry Claude route, or a MiniMax `/anthropic` proxy — `api_mode` is auto-detected as `anthropic_messages` so the subagent uses the right wire format without you setting anything. Set `api_mode` explicitly when the auto-detection guess is wrong (rare).
:::tip
The agent handles delegation automatically based on the task complexity. You don't need to explicitly ask it to delegate — it will do so when it makes sense.
:::