Feat: deepagent subagents#665
Closed
dylan-savage wants to merge 3 commits into
Closed
Conversation
Introduces agent_deepagent_subagent node type so DeepAgent orchestrators can delegate work to specialized sub-agents via the deepagent invoke channel, with per-subagent tools and LLM routing through its own host services. - services.subagent.json defines the new node type with description, system_prompt, and instructions config fields - IInvokeDeepagent (Describe/DescribeResponse) added to rocketlib.types for the describe fan-out protocol - DeepAgentDriver._collect_subagents builds deepagents.SubAgent entries from connected subagents, wiring each to its own LLM/tools host - JSON-schema args serialization so the LLM sees actual field names for tools like task, not opaque <class 'X'> strings - Stop word and tolerant first-object parser for the tool-call envelope to survive LLM outputs that emit duplicate or hallucinated JSON objects Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Both nodes get an Advanced Mode checkbox (crewai-style) that switches between Instructions and Agent Description + System Prompt - Subagent is now invoke-only: removed questions/answers lanes and narrowed classType to ["deepagent"] so it can only be reached from a manager via the deepagent channel - Driver cleanup: factor _compose_system_prompt helper used by both the manager and subagent paths, drop diagnostic logging, simplify parse and schema helpers Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
github-actions
Bot
added
module:server
|
No description provided. |
Contributor
📝 WalkthroughWalkthroughThis PR adds DeepAgent subagent support and an advanced-mode system prompt, selects between DeepAgent and DeepAgentSubagent drivers at startup, implements a Changes
Sequence Diagram(s)sequenceDiagram
participant Manager as Top-level DeepAgent Driver
participant Orchestrator as IInstance / IGlobal
participant SubagentNode as Subagent Node (agent_deepagent_subagent)
participant Rocketlib as Control-plane (IInvokeDeepagent)
Manager->>Orchestrator: send `deepagent.describe` (IInvokeDeepagent.Describe)
Orchestrator->>SubagentNode: receive invoke, _handle_deepagent_describe
SubagentNode-->>Orchestrator: return DescribeResponse (name, description, system_prompt, instructions, node_id, invoke)
Orchestrator-->>Manager: aggregate responses, emit SSE/debug, create_deep_agent(..., subagents=...)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
dylan-savage
requested review from
Rod-Christensen,
asclearuc and
ryan-t-christensen
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@nodes/src/nodes/agent_deepagent/deepagent.py`:
- Around line 670-673: The except block swallowing errors from
self._invoker.instance.invoke('deepagent', req, nodeId=node_id) should be
changed to log the exception at debug level; catch Exception as e and call the
node's logger (e.g. self.logger.debug or a module logger) to record a
descriptive message including the sub-agent name 'deepagent', the node_id, any
relevant parts of req, and the exception details/stacktrace so
connectivity/protocol failures during sub-agent discovery are visible for
troubleshooting.
In `@nodes/src/nodes/agent_deepagent/IInstance.py`:
- Around line 82-84: The branch in the invoke method that checks `isinstance(op,
str) and op == 'deepagent.describe'` currently calls
`self._handle_deepagent_describe(param)` and falls through implicitly; update
that branch to explicitly return None (e.g., `return None`) after calling
`_handle_deepagent_describe` so `invoke`'s return is explicit and satisfies
linters and readers.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: 82dab710-4063-4f0f-b3ed-d850c8e719d2
📒 Files selected for processing (7)
nodes/src/nodes/agent_deepagent/IGlobal.pynodes/src/nodes/agent_deepagent/IInstance.pynodes/src/nodes/agent_deepagent/deepagent.pynodes/src/nodes/agent_deepagent/services.jsonnodes/src/nodes/agent_deepagent/services.subagent.jsonpackages/server/engine-lib/rocketlib-python/lib/rocketlib/__init__.pypackages/server/engine-lib/rocketlib-python/lib/rocketlib/types.py
… parser The first-object JSON parser handles every case the stop-word caught. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@nodes/src/nodes/agent_deepagent/deepagent.py`:
- Around line 974-977: In the loop that iterates "for inst in instructions or
[]" avoid reassigning the loop variable "inst"; instead create a new name (e.g.,
inst_clean or inst_stripped) and use that in the if check and when appending to
"prompt" so replace "inst = inst.strip()" with a new variable and update the
conditional and f-string to use it, keeping the rest of the "prompt"
concatenation logic unchanged.
- Around line 519-521: Remove the redundant "from rocketlib import debug as
_debug" and update the debug call inside _run to use the already-imported debug;
specifically, delete the alias import and change the log expression that
references _debug (the f-string reporting tool_descriptors in function _run) to
call debug(...) so the module consistently uses the existing debug symbol (also
used elsewhere later in the file).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: 6cc065b9-abbf-4964-bb7f-f7606e720749
📒 Files selected for processing (1)
nodes/src/nodes/agent_deepagent/deepagent.py
Comment on lines
+519
to
+521
| from rocketlib import debug as _debug | ||
|
|
||
| _debug(f'deepagent _run discovered {len(tool_descriptors)} tools: {[td.get("name") if hasattr(td, "get") else str(td) for td in tool_descriptors]}') |
Contributor
There was a problem hiding this comment.
🧹 Nitpick | 🔵 Trivial
Remove redundant debug import.
debug is already imported at line 34. This re-import as _debug is unnecessary and creates naming inconsistency—line 566 uses debug directly while line 521 uses _debug.
♻️ Proposed fix
- from rocketlib import debug as _debug
-
- _debug(f'deepagent _run discovered {len(tool_descriptors)} tools: {[td.get("name") if hasattr(td, "get") else str(td) for td in tool_descriptors]}')
+ debug(f'deepagent _run discovered {len(tool_descriptors)} tools: {[td.get("name") if hasattr(td, "get") else str(td) for td in tool_descriptors]}')🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@nodes/src/nodes/agent_deepagent/deepagent.py` around lines 519 - 521, Remove
the redundant "from rocketlib import debug as _debug" and update the debug call
inside _run to use the already-imported debug; specifically, delete the alias
import and change the log expression that references _debug (the f-string
reporting tool_descriptors in function _run) to call debug(...) so the module
consistently uses the existing debug symbol (also used elsewhere later in the
file).
Comment on lines
+974
to
+977
| for inst in instructions or []: | ||
| inst = inst.strip() | ||
| if inst: | ||
| prompt = f'{prompt}\n{inst}' |
Contributor
There was a problem hiding this comment.
🧹 Nitpick | 🔵 Trivial
Minor: Avoid overwriting loop variable.
The assignment inst = inst.strip() shadows the loop variable, which can be confusing. Consider using a different name for clarity.
♻️ Proposed fix
for inst in instructions or []:
- inst = inst.strip()
- if inst:
- prompt = f'{prompt}\n{inst}'
+ stripped = inst.strip()
+ if stripped:
+ prompt = f'{prompt}\n{stripped}'📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| for inst in instructions or []: | |
| inst = inst.strip() | |
| if inst: | |
| prompt = f'{prompt}\n{inst}' | |
| for inst in instructions or []: | |
| stripped = inst.strip() | |
| if stripped: | |
| prompt = f'{prompt}\n{stripped}' |
🧰 Tools
🪛 Ruff (0.15.9)
[warning] 975-975: for loop variable inst overwritten by assignment target
(PLW2901)
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@nodes/src/nodes/agent_deepagent/deepagent.py` around lines 974 - 977, In the
loop that iterates "for inst in instructions or []" avoid reassigning the loop
variable "inst"; instead create a new name (e.g., inst_clean or inst_stripped)
and use that in the if check and when appending to "prompt" so replace "inst =
inst.strip()" with a new variable and update the conditional and f-string to use
it, keeping the rest of the "prompt" concatenation logic unchanged.
Collaborator
Author
|
These changes are going to be merged into feat/base-ui instead of develop |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Type
Testing
./builder testpassesChecklist
Linked Issue
Fixes #664
Summary by CodeRabbit
New Features
Bug Fixes