Deep Review

Quickstart

Review a directory and print the consolidated report. DeepReviewAgentSDKWorkflow.execute is an async coroutine, so drive it with asyncio.run (or await it inside an existing event loop):

import asyncio

from attune.workflows import DeepReviewAgentSDKWorkflow


async def main() -> None:
    workflow = DeepReviewAgentSDKWorkflow()
    result = await workflow.execute(path="src/", depth="standard")

    print(result.success)          # True on a completed review
    print(result.summary)          # short health summary
    print(result.final_output)     # the full consolidated report


asyncio.run(main())

depth defaults to "standard", so execute(path="src/") is equivalent. Use "quick" for a fast pass or "deep" for the fullest review.

Tasks

Review a path from the CLI

Goal: run a one-off review over a directory without writing any Python.

Steps:

# Default depth (standard) over a directory:
attune workflow run deep-review --path src/

# Deep review, JSON output for a pre-merge gate:
attune workflow run deep-review --path src/ --depth deep --json

# Narrow to the security pass via JSON input:
attune workflow run deep-review --path src/ --input '{"focus": ["security"]}'

Verify: --path / -p defaults to the current directory; --depth accepts quick, standard, or deep; --json / -j emits machine-readable output. There is no dedicated --focus flag — pass focus through --input as JSON. Use attune workflow info deep-review to confirm registration and attune workflow list to see it alongside the other workflows.

Call the review from Python

Goal: drive deep-review from a hook or pre-merge gate and act on the result.

Steps:

import asyncio

from attune.workflows import DeepReviewAgentSDKWorkflow


async def main() -> None:
    workflow = DeepReviewAgentSDKWorkflow()
    result = await workflow.execute(path="src/api/", depth="deep")

    if not result.success:
        print("review failed:", result.error)
        return

    print(result.final_output)
    for action in result.suggestions:
        print(action)


asyncio.run(main())

Verify: execute is a coroutine — await it. A completed review returns success=True with the report in final_output; a failure returns success=False with a populated error and error_type. metadata echoes the path, depth, max_turns, and active focus.

Run only the passes you need

Goal: scope the review to one or two domains instead of all three.

Steps:

import asyncio

from attune.workflows import DeepReviewAgentSDKWorkflow


async def main() -> None:
    workflow = DeepReviewAgentSDKWorkflow()

    # Security pass only:
    sec = await workflow.execute(path="src/auth/", focus=["security"])
    print(sec.final_output)

    # Security + quality, skip test gaps:
    both = await workflow.execute(
        path="src/auth/", focus=["security", "quality"]
    )
    print(both.final_output)


asyncio.run(main())

Verify: focus accepts any subset of "security", "quality", "test-gaps". Only the named passes run, and metadata["focus"] reflects the active set. An all-invalid focus returns success=False with an "Invalid focus values" error naming the valid options.

Reference

Deep-review's public surface is the DeepReviewAgentSDKWorkflow class, re-exported from attune.workflows. WorkflowResult comes from attune.workflows as well.

DeepReviewAgentSDKWorkflow — attune.workflows.deep_review

Symbol	Purpose
DeepReviewAgentSDKWorkflow()	Construct the workflow. Takes no special constructor arguments (no system_prompt_suffix).
DeepReviewAgentSDKWorkflow.execute(**kwargs)	Async. Run the review. Honors path (str, required), depth ("quick" / "standard" / "deep", default "standard"), and focus (list of "security" / "quality" / "test-gaps", default all three). Returns a WorkflowResult.
DeepReviewAgentSDKWorkflow.name	The registered slug, "deep-review".
DeepReviewAgentSDKWorkflow.stages	["deep-review"]; the stage runs at the CAPABLE model tier.

Depth → agent-turn budget

Depth	Max turns	Use when
quick	15	A fast pass on a small path.
standard	30	The default — balanced coverage and cost.
deep	50	The fullest review of a large or critical area.

The three passes

focus value	Subagent	Domain
security	security-reviewer	Injection, secrets, path traversal, auth, OWASP Top 10.
quality	quality-reviewer	Complexity, broad excepts, dead code, naming, duplication, type hints, docstrings, long functions.
test-gaps	test-gap-reviewer	Untested paths, missing edge cases, weak assertions, mocks hiding bugs.

WorkflowResult fields read after a review

Field	Type	Meaning
success	bool	Whether the review completed.
final_output	Any	The consolidated report — a serialized report when findings parse, else the raw markdown.
summary	str \| None	Short health summary.
suggestions	list[NextAction]	Prioritized next actions.
cost_report	CostReport	Cost / usage for the run.
provider	str	The provider that served the run.
metadata	dict	Echoes path, depth, max_turns, focus, and workflow; carries SDK error fields on failure.
error / error_type	str \| None	Failure reason and category ("config" / "runtime" / "provider" / "timeout" / "validation").

Entry points

Surface	Invocation
Skill	/deep-review in a Claude Code conversation.
CLI	attune workflow run deep-review --path <p> [--depth quick\|standard\|deep] [--json] [--input '{"focus": [...]}'].
MCP tool	deep_review — one required path argument; runs all three passes at standard depth (the handler does not pass depth or focus).
Python	await DeepReviewAgentSDKWorkflow().execute(path=<p>, depth=<d>, focus=[...]).