Release Prep

Quickstart

Run the gate from the CLI at your project root:

attune workflow run release-gate

The output is the readiness report — the verdict (APPROVED / BLOCKED), the four quality gates with actual-vs-threshold values, a per-agent breakdown, and any blockers or warnings. The canonical slug release-prep runs the same workflow.

Tasks

Gate a release from the CLI

Goal: get a pass/fail verdict before tagging a release.

Steps:

# Run the gate at the project root:
attune workflow run release-gate

# JSON output for a CI step:
attune workflow run release-gate --json

Verify: the slugs are release-gate and release-prep (both run ReleasePrepTeamWorkflow). --path / -p defaults to the current directory; --json / -j emits machine-readable output. The verdict is APPROVED or BLOCKED; the run exits 0 even when BLOCKED (the verdict is data, not the exit code).

Run the gate from Python

Goal: drive the gate from a release script and branch on the verdict.

Steps:

import asyncio

from attune.agents.release import ReleasePrepTeamWorkflow


async def main() -> None:
    workflow = ReleasePrepTeamWorkflow()
    result = await workflow.execute(path=".")

    approved = result.metadata["approved"]
    print("APPROVED" if approved else "BLOCKED", result.metadata["confidence"])
    print(result.summary)


asyncio.run(main())

Verify: execute is a coroutine — await it. result.success is True whenever the assessment ran; the verdict is result.metadata["approved"]. The full report is in result.final_output.

Customize the quality-gate thresholds

Goal: raise (or lower) the bars the gate enforces.

Steps:

import asyncio

from attune.agents.release import ReleasePrepTeam


async def main() -> None:
    team = ReleasePrepTeam(
        quality_gates={"min_coverage": 90.0, "min_quality_score": 8.0},
    )
    report = await team.assess_readiness(codebase_path=".")
    print(report.format_console_output())


asyncio.run(main())

Verify: assess_readiness is a coroutine — await it. The quality_gates keys are max_critical_issues, min_coverage, min_quality_score, and min_doc_coverage; any you pass override the defaults, the rest stay at DEFAULT_QUALITY_GATES.

Reference

Release-prep's public surface is the ReleasePrepTeamWorkflow (CLI / registry adapter) and the ReleasePrepTeam coordinator, both importable from attune.agents.release.

ReleasePrepTeamWorkflow — attune.agents.release

Symbol	Purpose
ReleasePrepTeamWorkflow(quality_gates=None, **kwargs)	Construct the registry adapter. quality_gates overrides thresholds.
ReleasePrepTeamWorkflow.execute(path=".", context=None, **kwargs)	Async. Run the gate. Maps target → path for CLI/VSCode. Returns a WorkflowResult.
ReleasePrepTeamWorkflow.name	The canonical slug, "release-prep" (synonym release-gate).
ReleasePrepTeamWorkflow.stages	["triage", "parallel-validation", "synthesis", "decision"].

ReleasePrepTeam — attune.agents.release

Symbol	Purpose
ReleasePrepTeam(quality_gates=None, redis_url=None)	Construct the coordinator. Optional Redis URL for coordination (graceful no-op when unavailable).
ReleasePrepTeam.assess_readiness(codebase_path=".")	Async. Run the four agents in parallel and return a ReleaseReadinessReport.
ReleasePrepTeam.get_total_cost()	Total LLM cost across agents ($0 in the default rule-based mode).

Default quality gates

Gate	Key	Default	Critical
Security	max_critical_issues	0	Yes
Test Coverage	min_coverage	80.0	Yes
Code Quality	min_quality_score	7.0	Yes
Documentation	min_doc_coverage	80.0	No

The four agents

Agent	Tool	Score basis
SecurityAuditorAgent	bandit (JSON, severity ≥ medium)	Severity-weighted; critical_issues = CRITICAL + HIGH.
TestCoverageAgent	pytest collect + pytest --cov	TOTAL coverage %; heuristic estimate as fallback.
CodeQualityAgent	ruff check --statistics	0–10 by violation count.
DocumentationAgent	AST docstring walk	Public-function docstring coverage %.

WorkflowResult fields read after a run

Field	Type	Meaning
success	bool	Whether the assessment ran (always True on a completed run — not the verdict).
final_output	dict	The serialized WorkflowReport (verdict callout, gate table, per-agent breakdown, blockers, warnings, next steps).
summary	str	Executive summary — approval status and the failed gates.
metadata	dict	approved (bool) and confidence (high / medium / low).

Entry points

Surface	Invocation
CLI	attune workflow run release-gate [--path <p>] [--json] (canonical slug release-prep).
Python	await ReleasePrepTeamWorkflow().execute(path=<p>) or await ReleasePrepTeam().assess_readiness(codebase_path=<p>).

There is no MCP tool for the gate — it is CLI / Python only.