github-atlassian
Advanced practitioner reference for GitHub and the Atlassian suite (Jira, Confluence, JSM, Bitbucket) covering 2025–2026 changes: GitHub rulesets, merge queue, GHAS rebundling (Secret Protection $19/Code Security $30), Copilot agent GA, AI Credits billing, artifact attestations, EMU; Atlassian Server EOL, Rovo bundling, JSM Service Collection, Forge replacing Connect, Data Center countdown; and the GitHub↔Jira integration pattern for 50–500 engineer orgs. Use when advising on GitHub governance, Atlassian configuration, tool selection, billing strategy, or cross-platform integration.
- License 许可
- MIT
- 最近更新
- 2026-11-15
- Stars
- ★ 5
- Forks
- 3
01 前言 Front Matter
- name
- github-atlassian
- description
- Advanced practitioner reference for GitHub and the Atlassian suite (Jira, Confluence, JSM, Bitbucket) covering 2025–2026 changes: GitHub rulesets, merge queue, GHAS rebundling (Secret Protection $19/Code Security $30), Copilot agent GA, AI Credits billing, artifact attestations, EMU; Atlassian Server EOL, Rovo bundling, JSM Service Collection, Forge replacing Connect, Data Center countdown; and the GitHub↔Jira integration pattern for 50–500 engineer orgs. Use when advising on GitHub governance, Atlassian configuration, tool selection, billing strategy, or cross-platform integration.
03 SKILL.md 正文
GitHub & Atlassian: Advanced Practitioner Reference (2026)
The Three Things You Must Plan Around Now
- GitHub billing changes (April–June 2026): GHAS split into two SKUs (April 1, 2025); Copilot moved to usage-based AI Credits billing (June 1, 2026)
- Bitbucket app-password hard deadline: brownouts start June 9, 2026; permanent removal July 28, 2026
- Atlassian Server is dead (Feb 2024); Data Center is on a countdown — no new DC sales after March 30, 2026; full EOL March 28, 2029
Part 1 — GitHub
1.1 Repositories and Governance
Rulesets have superseded branch protection as the model of choice:
- Multiple rulesets aggregate (most restrictive applies) vs branch protection (one rule "wins" by priority)
- Can be set at the organization level (GitHub Enterprise)
- Target branch-name patterns and tags
- Run in evaluate (dry-run) mode before enforcement
- Enforce metadata rules (commit message, branch naming, author email) natively — eliminating pre-receive hooks
- Branch protection still exists and layers with rulesets
Three repository visibilities:
- Public, private, and internal (enterprise accounts only) — internal = visible to all enterprise members, ideal for InnerSource
CODEOWNERS: routes review by path (last matching pattern wins); can be required via rulesets.
GitHub merge queue (GA 2023):
- Builds speculative merge commits (base + PRs ahead + this PR) and merges FIFO once required checks pass
- Requires the
merge_groupworkflow trigger — the single most common misconfiguration - Does not support wildcard branch patterns
- Designed and tested at scale (GitHub.com: 30,000+ PRs, 4.5M CI runs before GA)
1.2 Issues and Projects
YAML issue forms have largely superseded markdown templates for serious intake — structured dropdowns/checkboxes/validations.
GitHub Issues vs Jira decision:
- GitHub Issues + Projects: free, markdown-native, zero context-switch — sufficient for teams under ~20–30 developers
- Jira wins on: workflow enforcement (conditions/validators/post-functions), velocity/burndown/CFD reporting, cross-project portfolio visibility, issue hierarchy (Initiative→Epic→Story→Sub-task), 5,000+ Marketplace
GitHub Projects (v2) honest gaps vs Jira:
- No native velocity/cycle-time analytics
- Flat hierarchy (no true epics/parent-child without task-list workarounds)
- Limited fine-grained permissions ("QA can only move to Done" isn't expressible)
- Managed via GraphQL API
1.3 GitHub Actions
Core building blocks: triggers, matrix builds, concurrency controls, reusable workflows (workflow_call) vs composite/JS/Docker actions.
Security and scale features:
- OIDC federation: bind to Environments (
environment:<env>subject) with required reviewers, not just branches — the right way to gate production - Environments + deployment protection rules: required reviewers, wait timers, branch restrictions
- Actions Runner Controller (ARC): Kubernetes-native autoscaling using ephemeral runner pods (fresh pod per job, destroyed after — prevents state leakage and ensures no job is assigned to a shutting-down runner)
- ARC 0.14.0 (March 2026): moved to
actions/scalesetlibrary client - Earlier releases added: native sidecar DinD, automatic pod retries (up to 5), OpenShift support, Azure Key Vault secret retrieval (GA), dual-stack IPv4/IPv6
- ARC 0.14.0 (March 2026): moved to
Security best practices:
- Pin third-party actions to commit SHAs (Dependabot updates them)
- Minimize
GITHUB_TOKENpermissions; set explicitpermissionsblocks - Marketplace actions introduce third-party supply-chain risk
Repository secrets/variables scope: org, repo, and environment levels.
1.4 GitHub Advanced Security (GHAS) — Repriced April 1, 2025
GHAS was unbundled into two metered, committer-priced SKUs available even on Team:
| SKU | Price | Includes |
|---|---|---|
| Secret Protection | $19/active committer/mo | Push protection, validity checks, AI-detected generic secrets, 200+ partner patterns |
| Code Security | $30/active committer/mo | CodeQL code scanning, Copilot Autofix, security campaigns, dependency review |
Active committer definition: unique active committers on a 90-day window over repos where the feature is enabled.
Cost control: scope to specific repos — org-wide enablement counts every contractor and bot that pushes.
Copilot Autofix: covers 90%+ of alert types in JS/TS/Java/Python; suggestions remediate 2/3+ of found vulnerabilities.
Other security features:
- Dependabot: alerts, security updates, version updates, grouped updates via
dependabot.yml - Secret scanning push protection: blocks at push; delegated bypass (approval workflow); validity checks against provider APIs
- Custom patterns (up to 500/org via Hyperscan regex)
- Artifact attestations (GA June 2024):
actions/attest-build-provenancevia Sigstore → SLSA Build Level 2 out of the box, Level 3 with reusable workflows; no key management; Kubernetes admission controller for enforcement - Security campaigns (2025): org-level security debt remediation workflows
1.5 GitHub Copilot
The product is now a family of agents:
| Feature | Status | Notes |
|---|---|---|
| Code completions | GA | Unlimited and unbilled on paid plans |
| Copilot Chat (IDE + GitHub.com) | GA | PR summaries, repo Q&A, slash commands, @workspace |
| Copilot code review | GA (April 4, 2025) | Automatic (via rules) + on-demand; 1M+ devs in first month of preview |
| Copilot coding agent | GA (Sept 25, 2025) | Assign issue → opens draft PR via GitHub Actions |
| Copilot CLI | GA | — |
| Agent mode (VS Code, JetBrains) | GA by early 2026 | Multi-file/multi-step agentic sessions |
Billing changed June 1, 2026 to usage-based AI Credits (1 credit = $0.01, priced by token/model):
| Plan | Price | AI Credits included |
|---|---|---|
| Business | $19/user/mo | $19 credits |
| Enterprise | $39/user/mo | $39 credits (also requires GHEC ~$21/user → real cost ~$60/user) |
- Code completions remain unlimited and unbilled
- Copilot code review additionally consumes GitHub Actions minutes on private repos starting June 1, 2026 (public repos remain free)
- Agentic sessions can consume credits and Actions minutes quickly — set spend alerts
Honest assessment: measurably helps on boilerplate, tests, and low-to-medium-complexity well-tested codebases; underdelivers on novel architecture and large unfamiliar codebases.
1.6 Enterprise Identity (EMU)
True SCIM provisioning requires GitHub Enterprise Cloud with Enterprise Managed Users (EMU).
- Standard GHEC: only SCIM-triggered invitations (not full provisioning)
- EMU hard tradeoff: managed users cannot contribute to public repos or external OSS
- Okta + Entra split-IdP combination is explicitly unsupported
- Data residency (GHE.com) requires EMU
GitHub Enterprise options:
- GHEC: cloud-hosted, EMU optional, GitHub Connect bridge
- GHES: self-hosted HA (primary-replica, backup-utils), GitHub Connect bridge to GHEC; air-gapped option
1.7 Packages and Ecosystem
GHCR for GitHub-native flows; choose Azure Artifacts/Artifactory for polyglot enterprise artifact governance.
GitHub Apps vs OAuth Apps vs PATs vs Actions tokens:
- GitHub Apps (JWT→installation tokens): the right credential for automation
- PATs: simpler but coarser; fine-grained PATs now available
- Actions
GITHUB_TOKEN: repo-scoped, short-lived — prefer over PATs in Actions workflows
Codespaces: devcontainers + prebuilds; billed per compute-hour.
Part 2 — Atlassian
2.1 Server EOL and Data Center Countdown
| Milestone | Date |
|---|---|
| Server end of support | February 15, 2024 — DONE |
| No new DC sales | March 30, 2026 |
| DC license/tier/app purchases stop | March 30, 2028 |
| DC full EOL (read-only) | March 28, 2029 |
Migration path: Cloud Migration Assistant; "Atlassian Ascend"/FastShift programs for 1,000+ user orgs. A Bitbucket Hybrid License (DC + Cloud) arrives mid-2026.
2.2 Jira Software
Two project types:
- Team-managed (formerly next-gen): simple, autonomous
- Company-managed (classic): deep scheme model (workflow, permission, notification, issue type, field, screen schemes + project roles) — required for workflow gates, cross-project portfolio
Hierarchy: Initiatives (in Plans) → Epics → Stories/Tasks → Sub-tasks
Workflows: statuses/transitions with conditions, validators, post-functions — the feature that makes Jira worth it over GitHub Issues.
JQL tips: write index-friendly queries; avoid full-text scans at scale; use functions (currentUser(), membersOf(), startOfWeek()).
Jira Automation: no-code triggers/conditions/actions/branches at project and org level.
Jira Plans (Advanced Roadmaps): cross-team/program planning and dependency tracking.
2.3 Jira Service Management (JSM) — Now "Service Collection"
JSM + Assets + Rovo + Customer Service Management are now sold as the Service Collection.
Premium tier (~$47–53/agent/mo) gates the features that matter:
- CMDB/Assets
- Change management with CI/CD deployment gating
- Major incident management
- Atlassian Intelligence
- Virtual Service Agent
Billed per agent — requesters are free.
Opsgenie status: new sales ended June 4, 2025; full shutdown April 5, 2027; capabilities folded into JSM.
JSM vs ServiceNow vs Zendesk:
- JSM wins: Atlassian-ecosystem integration, per-agent price
- ServiceNow wins: heavyweight enterprise ITSM breadth
- Zendesk wins: pure customer-facing support UX
2.4 Rovo — Atlassian's AI Layer
"Rovo for all" (from April 9, 2025): Rovo Search/Chat/Agents/Studio included in paid Jira/Confluence/JSM Cloud plans — no separate SKU for core features.
Credit allowances (pooled at org level, monthly, no rollover):
| Plan | Credits/user/mo |
|---|---|
| Standard | 25 |
| Premium | 70 |
| Enterprise | 150 |
- Jira, Confluence, and Service Collection/JSM
- A single Rovo Chat/Agent question = 10 credits; Deep Research = 100 credits
- Atlassian Intelligence (in-product drafting/summaries) is the foundation; Rovo is the cross-product layer
- Rovo Dev (for engineers — CLI, PR analysis) is a separate $20/dev/mo product with 2,000 credits
- Rovo is Cloud-only
- Rovo credit overages are not yet billed but Atlassian has signaled future consumption-based charging (≥90 days notice)
2.5 Bitbucket — URGENT Deprecation
App-password hard deadline:
- No new creation: September 9, 2025 — PASSED
- Controlled brownouts begin: June 9, 2026
- Permanent removal: July 28, 2026
Action required immediately: migrate all Bitbucket CI integrations to API tokens with scopes (or OIDC for cloud auth).
Bitbucket Pipelines OIDC: oidc: true flag; $BITBUCKET_STEP_OIDC_TOKEN for keyless cloud auth.
Code Insights/Annotations API: pushes CI quality/coverage/security results back onto PRs.
Bitbucket Pipelines vs GitHub Actions:
- Actions wins on: ecosystem (20,000+ marketplace), reusable workflows, matrix builds, macOS/Windows runners, free minutes (2,000/mo vs Bitbucket's 50 free; Standard 2,500, Premium 3,500)
- Pipelines wins on: simplicity and zero-config Jira deployment tracking
Main reason to keep Bitbucket: native Jira integration — same-vendor, deeper integration than GitHub's, strongest reason Atlassian-centric orgs stay
2.6 Confluence
Key capabilities:
- Spaces with a permission model (space admin; page view/add/edit/delete/export)
- Page hierarchy and restrictions
- Fabric editor with macros (code, ToC, panels, status, live Jira issue lists, children display, excerpt/excerpt-include)
- Whiteboards (GA February 29, 2024): Cloud-only
- Databases (broad rollout July 10, 2024): structured data as fields + entries, sync from Jira/Confluence/CSV — Cloud-only
- CQL search and Confluence Analytics (Cloud)
- Page properties + reports for lightweight tracking
Confluence vs Notion vs SharePoint:
- Confluence wins: Atlassian-integrated teams (live Jira embeds, JSM knowledge base, space permissions)
- Notion wins: flexible databases/UX for smaller teams
- SharePoint wins: Microsoft 365-centric document governance
2.7 Forge vs Connect
Forge has replaced Connect as the app platform.
| Connect | Forge | |
|---|---|---|
| Model | External iframe (third-party servers) | Serverless on Atlassian infrastructure |
| Storage | External DB | Key-value + Forge SQL |
| Compliance | Requires vendor trust | Preferred model (Atlassian-hosted) |
| Status | End of support Q4 2026 | Active, all new investment |
Connect deprecation timeline:
- No new Connect listings: September 17, 2025 — PASSED
- No Connect updates: March 31, 2026
- End of support: Q4 2026
Action required: audit Connect app dependencies; confirm vendors have Forge versions before Q4 2026.
2.8 Atlassian Guard (Security Layer)
- Guard Standard (SSO/SCIM/API-token control/audit log) — included free in Cloud Enterprise
- Guard Premium (data classification, content scanning/DLP, anomalous-activity threat detection, extended audit) — currently supports Jira and Confluence
Part 3 — GitHub ↔ Atlassian Integration
The Official "GitHub for Jira" App
- Atlassian-maintained; Jira Cloud only
- Links commits/branches/PRs/builds/deployments to issues by string-matching the Jira issue key
- Surfaces data in the Jira development panel, board card icons, Code/Releases/Deployments tabs, JQL
- Smart commits:
PROJ-123 #comment ... #time 2h #done - "Create branch" button in Jira; JQL queries over development data
Limitation: only as reliable as naming discipline — which is why mature teams enforce issue-keyed branch names via rulesets and drive status transitions via Actions.
GitHub Enterprise Server ↔ Jira DC: not supported by the native app — requires third-party (e.g., GitKraken's Git Integration for Jira).
Production Integration Pattern (50–500 Engineers)
- Enforce issue-keyed branch names via GitHub rulesets (regex rejecting non-conforming branches at push)
- Drive Jira status transitions from GitHub Actions on
pull_request/push/deployment events calling the Jira API oratlassian/gajira-*actions — more reliable than smart commits (which depend on developer discipline and email matching) - Track deployments via GitHub Environments → Jira deployment panel
- Pipe CodeQL/Dependabot findings into Jira as security issues via Actions for SLA-tracked remediation
- Keep Jira as system of record; use GitHub Projects only for engineering-local sprint views to avoid dual-maintenance
Part 4 — Cross-Cutting Concerns
Org Design
- Mono-repo vs poly-repo drives GitHub org/team structure and Jira project layout (one project per team is the common default)
- Standardize naming across both: branch names, PR titles, Jira summaries
- Standardize identity: one IdP (Entra ID/Okta) → GitHub teams via EMU SCIM → Jira groups/roles via Guard SCIM
End-to-End Agile Delivery Flow
Jira issue → issue-keyed branch → PR (linked) → CI (Actions/Pipelines) → human + Copilot code review → merge queue → Jira transition → release → deployment tracked in Jira
- Dual-track agile: Confluence (discovery) + Jira sprints (delivery)
- DoD enforcement: Jira automation + GitHub required checks
DORA Metrics Cross-Platform
- Deployment frequency: GitHub releases/deployments
- Lead time: Jira create → PR merge → deploy
- Change failure rate: bug/incident tickets
- MTTR: incident resolution
- Tooling: LinearB, Sleuth, Faros AI, Jellyfish, Haystack
Security and Compliance Integration
- GHAS findings → Jira security issues (via Actions)
- Both audit logs → SIEM
- JSM change requests linked to GitHub PRs/deployments
- Secret-scanning alerts → JSM incidents
- Both platforms carry SOC 2 / ISO 27001; Atlassian Cloud adds FedRAMP
Licensing and Cost Optimization
| Product | Standard | Premium |
|---|---|---|
| Jira | ~$7.91/user/mo | ~$14.54/user/mo |
| Confluence | ~$5.42/user/mo | ~$10.44/user/mo |
| JSM | ~$20/agent/mo | ~$47–53/agent/mo |
| GitHub Team | $4/user/mo | — |
| GitHub Enterprise | ~$21/user/mo | — |
| GHAS Secret Protection | $19/committer/mo | — |
| GHAS Code Security | $30/committer/mo | — |
| Copilot Business | $19/user/mo | — |
| Copilot Enterprise | $39/user/mo (+ GHEC) | — |
Optimization levers:
- Prune inactive users (metered billing means unused seats still cost)
- Scope GHAS to specific repos, not org-wide
- Reconcile licenses against IdP group sync
- Consolidate on Enterprise where volume discounts apply
- Set Copilot spend alerts given usage-based billing
Staged Implementation Roadmap
Stage 1 — Stabilize Identity and Deprecation Exposure (Now)
- Migrate Bitbucket app passwords to API tokens/OIDC before June 9, 2026 brownouts — this is the most urgent hard deadline
- Audit Connect app dependencies; confirm vendors have Forge versions before Q4 2026
- Standardize SSO/SCIM on one IdP feeding both ecosystems
- Decide EMU vs standard GHEC using the OSS-contribution and data-residency tests
Stage 2 — Modernize Governance (Next Quarter)
- Migrate branch protection → organization rulesets in evaluate mode first, then enforce; add metadata rules for issue-keyed branch names
- Enable merge queue on busy default branches (remember the
merge_grouptrigger) - Scope GHAS Secret Protection + Code Security to active repos; turn on push protection org-wide with delegated bypass
Stage 3 — Integrate and Measure (6–12 Months)
- Deploy GitHub for Jira; replace smart-commit reliance with Actions-driven transitions
- Stand up DORA metrics via a cross-platform tool
- Pilot Copilot coding agent / code review and Rovo agents on contained workflows; set spend alerts
Thresholds That Change the Plan
- Cross ~30 engineers or need workflow gates/portfolio reporting → move planning to Jira (not GitHub Projects)
- Active-committer count balloons with bots/contractors → re-scope GHAS repos to control cost
- Copilot agentic usage drives credit overages → cap org spend and restrict premium models
- Air-gapped/sovereign requirements → GHES + Bitbucket DC (until 2029)
04 同仓库更多 Skills
white-paper-writing
Use when writing white papers, technical reports, policy documents, thought leadership pieces, or any long-form authoritative document. Triggers on requests to write, structure, outline, or review white papers, technical reports, position papers, policy briefs, or evidence-driven arguments. Also use for B2B content marketing strategy, technical documentation requiring authority, academic-style reports, or any document that must educate while persuading. Use when someone asks "how should I structure this" for a long-form document.
prose-technical-mechanics
Use when editing or writing prose at a technical/mechanical level: choosing between Anglo-Saxon and Latinate diction, placing information in topic vs stress positions, writing periodic vs cumulative sentences, handling sentence rhythm, punctuation as logic, transitions, paragraph structure, or phonesthetics. Use when someone wants to understand WHY prose works or doesn't, not just surface-level style advice. Triggers on word choice, sentence structure, emphasis, rhythm, diction, or any deep writing mechanics question.
prose-mechanics
Use when writing, editing, or analyzing prose style; when trying to avoid generic AI-sounding writing; when applying specific prose styles such as journalistic, academic, narrative, KJV biblical, Hemingway plain, Latinate/Jamesian, Orwellian, Augustan, Victorian, Early Modern (Bacon, Browne, Milton), or free indirect discourse; or when coaching someone on writing voice and mechanics. Triggers on any request to improve, analyze, or write in a specific prose style, to fix "AI-sounding" writing, to make prose more vivid or concrete, to vary sentence rhythm, or to understand how a particular writer achieves their effects.
narrative-structure
Use when structuring stories, pitches, presentations, white papers, or any persuasive narrative; analyzing the hero's journey in fiction or non-fiction; applying StoryBrand for business communication; designing narrative arcs for thought leadership; or understanding Vogler, Snyder, Propp, or Campbell frameworks. Triggers on any narrative structure, storytelling, pitch structure, StoryBrand, hero's journey, three-act structure, Freytag, Save the Cat, narrative arc, or "how should I structure this story/pitch/presentation" request.
legal-brief-writing
Use when writing legal briefs, motions, or any court filing; when analyzing judicial cognition and persuasion; when applying elite brief-writing techniques from Roberts, Clement, Kagan, Garner; or when structuring legal arguments for maximum persuasion. Also use for autistic/neurodivergent advocates who need to understand how their natural communication style maps to elite legal advocacy. Triggers on any legal writing, brief drafting, motion preparation, or legal argument structuring request.
journal-publishability-criteria
Use whenever a research paper, article or preprint is being written, revised, reviewed or published, and specifically before every publication of docs/paper.md. Covers what legitimate journals actually gate on, in order of gating — the editor's desk screen (scope, novelty, package problems), what reviewers judge (methodological soundness, claims the evidence can carry, statistical and qualitative reporting), EQUATOR reporting guidelines, the ICMJE authorship criteria, COPE's view of text recycling, the near-universal generative-AI policy (no AI author, disclosure, author responsibility), TOP 2025 transparency standards, how to verify a journal is legitimate, submitting without an institution, and a pre-submission checklist. Triggers on requests to evaluate, review, prepare or submit a paper, to judge publishability, to write a declarations or data-availability section, to disclose AI use, or to check a journal.