AI GitHub 릴리스 블로그 라이터 · 자율 에이전트

AI GitHub 릴리스 블로그 라이터

An autonomous AI agent that scans your merged PRs, filters out internal noise, translates engineering language into user-facing copy, and drafts a clean changelog article grouped into What's new, Improved, Fixed, and Heads up — publish-ready every week.

연동: github · wordpress · gitlab · linear · slack · notion
live · ask github release blog writer
ai-github-release-blog-writer
Write this week's release post for flowhunt/web — kept user-facing changes only. alex · product · 9:14am
R AI GitHub Release Blog Writer · thinking · 3 tool calls · 1.4s
  • Scanning flowhunt/web · 47 PRs merged in 7 days
  • Filtering 19 internal changes (refactors, deps, CI)
  • Rewriting 28 user-facing changes in your voice
FlowHunt v3.42 — what shipped this week
draft · 612 words · 4 sections · cited 28 PRs
ready
What's new
  • Inline citations in chat replies — every claim links back to its source doc
  • Slack thread mode — keep context across replies, not just first messages
Improved
  • Knowledge base ingest is 4× faster on large PDFs
  • Agent memory now auto-compresses runs older than 30 days
Fixed
  • Editor crashed when pasting from Notion (regression in 3.41.2)
  • SSO timeouts on first login when MFA was required
Heads up
  • Legacy /v1/keys endpoint deprecated — switches off Dec 15
▸ ready for WordPress · publish · edit · regenerate
실제 동작 보기

병합된 PR에서 게시된 체인지로그까지, 단계별로.

One real example showing exactly what happens when you give the agent a task.

1 YOUR PROMPT
AP
Alex Park
Product Lead · DevStack Inc.
Write this week's release blog post. Cover everything merged since last Friday. Skip internal refactors and dependency bumps — focus on what users will notice.
just now
2 AGENT AT WORK running
GitHub Scanning merged PRs from the last 7 days
47 PRs found across 3 repos · date range confirmed
GitHub Filtering internal noise: refactors, dep bumps, CI fixes
19 internal PRs filtered · 28 user-facing PRs remain
Knowledge Base Loading product context and past changelog tone
voice profile loaded · feature naming conventions retrieved
LLM Draft Grouping PRs into What's new / Improved / Fixed / Heads up
4 new features · 9 improvements · 12 fixes · 3 breaking changes
LLM Draft Translating engineering descriptions to user-facing copy
28 entries rewritten · jargon removed · benefits highlighted
WordPress Publishing changelog post with formatted sections
published · v2.14 release notes live
Slack Notifying #product and #marketing with published link
2 channels notified · review link attached
3 THE OUTPUT ready for review
Changelog Article Draft Publish

DevStack v2.14 — What's New This Week

DevStack Blog·just now·Published to WordPress · devstack.io/blog/changelog

What's new:

- Real-time collaboration: Multiple users can now edit the same workflow simultaneously with live cursors and conflict resolution.

- Webhook triggers: Fire workflows from any external service via a unique webhook URL. No polling required.

Improved:

- Dashboard load time is 40% faster on workspaces with 500+ workflows.

- CSV export now handles Unicode characters correctly across all locales.

Fixed:

- Fixed an issue where scheduled runs would skip if the server clock drifted by more than 2 seconds.

- API rate limit headers now return accurate remaining counts.

Heads up:

- The legacy /v1/tasks endpoint will be removed on July 1. Migrate to /v2/tasks before then.

📄 16 paragraphs ·SEO score: ✅
▸ publish to WordPress · edit · regenerate
상시 릴리스 노트

체인지로그가 밀리는 일은 이제 없습니다.

The agent runs on a weekly schedule — scanning repos, filtering noise, drafting copy and publishing — all visible on the kanban. Your users get consistent release communication without anyone on your team touching a doc.

app.flowhunt.io/ai-projects/projects/ai-github-release-blog-writer/issues
라이브
GitHub Release Blog Writer · marketing-workspace

이슈

실행 중
열림 0
진행 중 0
완료 0
활성 에이전트
3 / 5
오늘 완료
8
사이클 시간
2h 14m
자동 해결
62%
작동 방식

GitHub 리포지토리에서 게시된 체인지로그까지 3단계.

Most users have their first release post drafted within ten minutes of connecting their repos.

01
리포지토리와 게시 대상을 연결하세요.
Link your GitHub repositories (or GitLab), then connect where you want changelogs published — WordPress, Notion, or Slack. Add your product context as a knowledge source so the agent understands feature names, user personas, and your changelog tone.
새 프로젝트
Publish weekly release notes from merged PRs.
GitHubWordPressProduct docs
초안 저장 에이전트 시작 →
02
에이전트가 스캔, 필터링, 초안을 작성합니다.
The agent scans all merged PRs since the last changelog, filters out internal noise (refactors, dependency bumps, CI changes), and groups the rest into user-facing categories. It translates engineering PR titles into benefits-oriented copy that your users actually understand.
This week's release
Scanned
47 PRs from 3 repos
Filtered
19 internal removed
Drafted
28 entries grouped
03
한 번 검토하고, 게시하고, 반복하세요.
The changelog pauses for your review. Make edits — every change trains the agent's memory for next time. Once approved, it publishes to WordPress and notifies your team on Slack. Set it on a weekly schedule and the agent handles the rest automatically.
Release history
Published
v2.14 · this week
Scheduled
v2.15 · next Fri
Archive
12 past releases
효과적인 이유

체인지로그가 자동으로 작성되면 무엇이 달라질까요.

01

Ship release notes the same day you ship code.

No more two-week lag between deploying features and telling users about them. The agent drafts your changelog within minutes of your release cut, so users discover new features while they are still fresh.

02

Engineering language becomes user-facing copy.

PR titles like 'refactor query planner to use CTE-based approach' become 'Dashboard loads 40% faster on large workspaces.' The agent understands your product context and writes for your users, not your engineers.

03

Consistent format, every single week.

FlowHunt agents have real memory. After a few release cycles, the agent has internalized your changelog structure, tone, and which types of changes to highlight versus skip. The format stays consistent even when different PMs run the release.

기능

PR 스캔부터 릴리스 게시물 게시까지, 모두 처리.

Six capabilities that replace the awkward weekly ritual of chasing engineers for release notes.

Automated PR scanning and classification
Scans all merged PRs across your connected repos. Classifies each as new feature, improvement, bug fix, breaking change, or internal-only — using PR labels, titles, descriptions, and linked issues.
User-facing copy from engineering context
Translates technical PR descriptions into benefits-oriented changelog entries. Pulls from your product knowledge base to use correct feature names, user personas, and consistent terminology.
Smart noise filtering
Automatically filters refactors, dependency bumps, CI pipeline changes, and internal tooling PRs. Configurable rules let you define what counts as internal versus user-facing for your team.
Scheduled weekly publishing
Runs on a trigger — every Friday, every sprint end, or on-demand. Drafts the changelog, queues it for review, and publishes to WordPress or Notion once approved.
Full observability on every run
Every PR scanned, every classification decision, every draft revision is logged and replayable. Drill into any run to see exactly which PRs were included, which were filtered, and why.
Approval gates for sensitive changes
Breaking changes and deprecation notices always flag for human review. Set rules for which categories require approval and which can auto-publish.
내부 구조

실제 에이전트 인프라 위에 구축되었습니다.

The Release Blog Writer runs on the same platform that powers 120+ autonomous agents. Open it in the visual editor and customize anything.

  • 비주얼 플로우 에디터. Every agent is a flow you can open and customize — swap the AI model, add filtering rules, wire in additional repos. No code.
  • 지식 소스 및 RAG. Connect product docs, past changelogs, feature naming guides, or user personas. The agent retrieves context before translating every PR.
  • 영구 메모리. Episodic, long-term and shared memory. The agent remembers your changelog tone, which PR types to skip, and how you rename features for users.
  • 멀티 에이전트 팀. Stack the release writer under a Supervisor alongside your content engine and social agents. The changelog feeds your newsletter and Twitter announcements automatically.
flowhunt · github-release-blog-writer · flow
AGENT
Release Blog Writer
S
PR Scanner
▸ 47개 PR 스캔
F
Filter
▸ 노이즈 제거
W
Writer
▸ 체인지로그 작성
연결된 도구
github wordpress slack notion linear web.search
메모리 · 634 이벤트
대상 사용자

빠르게 배포하고 더 빠르게 소통하는 팀을 위해 제작.

If you deploy weekly but your changelog is a month behind, this agent closes the gap.

PRODUCT
프로덕트 팀
Stop chasing engineers for release notes every Friday. The agent reads the PRs directly and drafts user-facing copy that product managers can review and ship in minutes.
DEVTOOLS
개발자 도구 기업
Your users read changelogs. Ship professional, consistent release notes that build trust and reduce support tickets about undocumented changes.
OSS
오픈소스 관리자
Keep your community informed without spending your weekends writing release notes. The agent handles the changelog so you can focus on code.
SAAS
B2B SaaS 기업
Turn your release cadence into a marketing asset. Consistent changelog posts show prospects that your product is actively developed and well-maintained.
AGENCY
개발 에이전시
Deliver professional release notes to clients after every sprint. The agent translates your team's work into language that non-technical stakeholders understand.
STARTUP
빌드 인 퍼블릭 스타트업
Share your shipping velocity with your audience. The agent turns your weekly PRs into engaging build-in-public content that attracts early adopters.
함께 사용하기 좋은 에이전트

프로덕트 팀의 나머지 에이전트와 연결하세요.

Stack agents under a Supervisor and let them hand work to each other — your release writer can feed the newsletter agent, the social agent, and the docs updater.

첫 번째 체인지로그, 몇 분 안에 초안 작성.

Start free. Connect your GitHub repos, point the agent at your last release, and watch it draft. No credit card required — cancel anytime.

자주 묻는 질문

What is an AI GitHub release blog writer?
An AI GitHub release blog writer is an autonomous agent that scans your merged pull requests, filters out internal changes, translates engineering language into user-facing copy, and drafts a structured changelog article. FlowHunt's version groups changes into What's new, Improved, Fixed, and Heads up sections — publish-ready for your blog or docs site.
How does the agent decide which PRs are user-facing?
The agent classifies PRs using a combination of labels, title patterns, description content, and linked issues. It filters out refactors, dependency bumps, CI changes, and internal tooling by default. You can customize these rules — for example, always include PRs labeled 'user-facing' and always skip PRs labeled 'internal'.
Can it work with GitLab or Linear instead of GitHub?
Yes. The agent supports GitHub and GitLab as source-of-truth for PRs and merge requests. It can also pull context from Linear issues to enrich the changelog with product context like feature names and epic descriptions.
How does it translate engineering language into user copy?
The agent uses your product knowledge base — feature naming conventions, user personas, past changelogs — to rewrite PR titles and descriptions in benefits-oriented language. 'Refactor query planner to use CTE-based approach' becomes 'Dashboard loads 40% faster on large workspaces.' Every edit you make trains its memory for next time.
Can I set it to run automatically every week?
Yes. FlowHunt supports triggers and scheduling. You can set the agent to run every Friday at 4pm, at the end of each sprint, or on-demand. It drafts the changelog and queues it for your review — or publishes automatically if you enable that.
How much does it cost per changelog?
FlowHunt uses a simple credit-based pricing model — each credit is worth about one dollar. A typical weekly changelog covering 20-50 PRs uses 1-3 credits depending on the number of PRs and depth of rewriting. You can try it free with trial credits.