Run any Skill in Manus with one click

$pwd:

sdk-release

Name: Sdk Release
Author: OpenHands

// This skill should be used when the user asks to "release the SDK", "prepare a release", "publish a new version", "cut a release", "do a release", or mentions the SDK release checklist or release process. Guides through the full software-agent-sdk release workflow from version bump to PyPI publication, emphasizing human checkpoints.

Run Skill in Manus

$ git log --oneline --stat

stars:744

forks:261

updated:May 4, 2026 at 14:05

File Explorer

2 files

SKILL.md

readonly

name	sdk-release
description	This skill should be used when the user asks to "release the SDK", "prepare a release", "publish a new version", "cut a release", "do a release", or mentions the SDK release checklist or release process. Guides through the full software-agent-sdk release workflow from version bump to PyPI publication, emphasizing human checkpoints.

SDK Release Guide

This skill walks through the software-agent-sdk release process step by step.

🚨 CRITICAL: NEVER merge the release PR or create/publish a GitHub release without the human's explicit approval. Release is the last line of human defense. Always present the current status and ask for confirmation before performing any irreversible action.

Phase 1: Trigger the Prepare-Release Workflow

Determine the target version (SemVer X.Y.Z). Then trigger the prepare-release.yml workflow, which creates a release branch and PR automatically.

Via GitHub UI

Navigate to https://github.com/OpenHands/software-agent-sdk/actions/workflows/prepare-release.yml, click Run workflow, enter the version (e.g. 1.16.0), and run it.

Via GitHub API

curl -X POST \
  -H "Authorization: token $GITHUB_TOKEN" \
  -H "Accept: application/vnd.github+json" \
  "https://api.github.com/repos/OpenHands/software-agent-sdk/actions/workflows/prepare-release.yml/dispatches" \
  -d '{
    "ref": "main",
    "inputs": {
      "version": "1.16.0"
    }
  }'

The workflow will:

Validate version format
Create branch rel-<version>
Run make set-package-version version=<version> across all packages
Update the sdk_ref default in the eval workflow
Open a PR titled "Release v<version>" with labels integration-test, behavior-test, and test-examples

⏸ Checkpoint — Confirm PR Created

Verify the PR exists and the version changes look correct before continuing.

gh pr list --repo OpenHands/software-agent-sdk \
  --head "rel-<version>" --json number,title,url

Phase 2: Address Deprecation Deadlines

The deprecation-check CI job runs on every PR. If the release version crosses any deprecation deadline declared in the codebase, the check will fail.

Review the failing check output and either:

Remove the deprecated code if the deadline has passed, or
Extend the deadline with justification.

Push fixes to the release branch. The check must pass before merging.

Phase 3: Wait for CI — Tests Must Pass

The release PR triggers three labeled test suites. All three must pass.

Label	Suite	What it covers
`integration-test`	Integration tests	End-to-end agent scenarios
`behavior-test`	Behavior tests	Agent behavioral guardrails
`test-examples`	Example tests	All runnable examples in `examples/`

Monitor status:

gh pr checks <PR_NUMBER> --repo OpenHands/software-agent-sdk

⏸ Checkpoint — Human Judgment on Failures

Some test failures may be pre-existing or flaky. Decide with the team whether each failure is:

Blocking — must fix before release
Known / pre-existing — acceptable to release with a follow-up issue
Flaky — re-run the workflow

Re-run failed jobs:

# Find the run ID
gh run list --repo OpenHands/software-agent-sdk \
  --branch "rel-<version>" --limit 5

# Re-run failed jobs
gh run rerun <RUN_ID> --repo OpenHands/software-agent-sdk --failed

Phase 4: Run Evaluation (Optional but Recommended)

Trigger an evaluation run on SWE-bench (or another benchmark) against the release branch to catch regressions. See the run-eval skill for full details.

curl -X POST \
  -H "Authorization: token $GITHUB_TOKEN" \
  -H "Accept: application/vnd.github+json" \
  "https://api.github.com/repos/OpenHands/software-agent-sdk/actions/workflows/run-eval.yml/dispatches" \
  -d '{
    "ref": "main",
    "inputs": {
      "benchmark": "swebench",
      "sdk_ref": "v<version>",
      "eval_limit": "50",
      "reason": "Pre-release eval for v<version>",
      "allow_unreleased_branches": "true"
    }
  }'

⏸ Checkpoint — Evaluate Results

Compare the eval results against the previous release. Significant score drops should block the release.

Phase 5: Merge the Release PR

🚨 STOP — Do NOT merge without explicit human approval. Present the CI status summary and ask the human to confirm before merging. Merging is effectively irreversible — it automatically triggers the full release pipeline (GitHub release → PyPI publish → downstream version bumps).

Once the human approves:

gh pr merge <PR_NUMBER> --repo OpenHands/software-agent-sdk --merge

Phase 6: Automated Release Pipeline (no action needed)

When the release PR is merged, the following happens automatically:

create-release.yml detects the merged rel-* branch, creates a GitHub release with tag v<version> and auto-generated release notes.
pypi-release.yml triggers on the published release and publishes all four packages to PyPI:
- openhands-sdk
- openhands-tools
- openhands-workspace
- openhands-agent-server
version-bump-prs.yml triggers after successful PyPI publish and creates downstream version bump PRs.

⏸ Checkpoint — Verify PyPI Publication

# Check each package is available (allow a few minutes for indexing)
for pkg in openhands-sdk openhands-tools openhands-workspace openhands-agent-server; do
  curl -s -o /dev/null -w "$pkg: %{http_code}\n" \
    "https://pypi.org/pypi/$pkg/<version>/json"
done

All should return 200.

Phase 7: Post-Release Announcements

After the automated pipeline completes, compose a Slack message for the human to post, including links to the downstream version bump PRs:

🚀 *SDK v<version> published to PyPI!*

Version bump PRs:
• <https://github.com/All-Hands-AI/OpenHands/pulls?q=is%3Apr+bump-sdk-<version>|OpenHands>
• <https://github.com/OpenHands/openhands-cli/pulls?q=is%3Apr+bump-sdk-<version>|OpenHands-CLI>

Release: <https://github.com/OpenHands/software-agent-sdk/releases/tag/v<version>|v<version>>

See references/post-release-checklist.md for details on reviewing downstream PRs and handling any issues.

Quick Reference — Full Checklist

related-skills.json

same repository

manage-evals.md

from "OpenHands/software-agent-sdk"

This skill should be used when the user asks to "trigger an eval", "run evaluation", "run swebench", "run gaia", "run benchmark", "compare eval runs", "compare evaluation results", "check eval regression", "compare benchmark results", "what changed in the eval", "diff eval runs", or mentions triggering, comparing, or reporting on SWE-bench, GAIA, or other benchmark evaluation results. Provides workflow for triggering evaluations on different benchmarks, finding and comparing runs, and reporting performance differences.

2026-05-11744

cross-repo-testing.md

from "OpenHands/software-agent-sdk"

This skill should be used when the user asks to "test a saas cross-repo feature", "deploy a feature branch to staging", "test SDK against OH Cloud branch", "e2e test a cloud workspace feature", "test secrets saas inheritance", or when changes span the SDK and OpenHands enterprise and need end-to-end validation against a staging deployment.

2026-04-05744

feature-release-rollout.md

from "OpenHands/software-agent-sdk"

This skill should be used when the user asks to "rollout a feature", "complete feature release", "propagate SDK feature", "track feature support", "what's missing for feature X", or mentions checking CLI/GUI/docs/blog support for SDK features. Guides agents through the multi-repository feature release workflow from SDK to docs to marketing.

2026-02-14744

debug-test-examples-workflow.md

from "OpenHands/software-agent-sdk"

Guide for debugging failing example tests in the `test-examples` labeled workflow. Use this skill when investigating CI failures in the run-examples.yml workflow, when example scripts fail to run correctly, when needing to isolate specific test failures, or when analyzing workflow logs and failure patterns.

2026-02-11744

code-style-guide.md

from "OpenHands/software-agent-sdk"

Project coding standards and style guidelines. Always follow these conventions when writing or reviewing code.

2026-01-08744

rot13-encryption.md

from "OpenHands/software-agent-sdk"

This skill helps encrypt and decrypt messages using ROT13 cipher. Use when the user asks to "encrypt" or "decrypt" a message.

2026-01-07744

package.json