Azure devops git
Skill smss123/azure-devops-git
skill and agent to make your ai strong in git , git lab, github,azure devops
npx -y skills add smss123/azure-devops-gitAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Expert Azure DevOps + Git skill for managing repositories, branches, pull requests, pipelines, work items, and long-running tasks via sub-agents. Use this skill whenever the user mentions: Azure DevOps, ADO, Azure Repos, Azure Pipelines, YAML pipelines, git in ADO context, PRs in DevOps, branch policies, work items, sprints, backlogs, boards, service connections, release pipelines, build agents, PAT tokens, secrets, variable groups, Key Vault, CODEOWNERS, coverage gates, changelogs, semver, policy drift, rollbacks, conflict resolution, or multi-repo changes. Also trigger when the user wants to: clone/push/pull from ADO repos, manage branches at scale, run parallel tasks, coordinate sub-agents for long jobs, automate CI/CD, manage ADO via REST API, implement git flow, use git worktrees, handle merge conflicts, revert bad releases, monitor agent health, order feature dependencies, auto-assign PR reviewers, generate release notes, send Teams or email alerts, detect policy drift, automate versioning, audit agent actions, coordinate atomic changes across multiple repos, kick off a full sprint with one command, use a Team Leader agent to manage sub-agents, route requests between agents, enhance prompts before agents run, share knowledge between agents, dynamically reassign failed tasks, or split oversized agent work. Do NOT skip this skill even if the request seems simple — it contains critical patterns and pitfalls.
SKILL.md
12.1 KB, as published. Nobody here has run it
author : samer abd allah (xprema systems)
Azure DevOps Git & Sub-Agent Orchestration Skill
Quick Reference — Read First
| Task Category | Jump To |
|---|---|
| Auth & Setup | Authentication |
| Git Operations | [references/git-ops.md] |
| Repos & Branches | [references/repos-branches.md] |
| Pull Requests | [references/pull-requests.md] |
| Pipelines (YAML/Classic) | [references/pipelines.md] |
| Work Items & Boards | [references/work-items.md] |
| REST API Patterns | [references/rest-api.md] |
| Git Flow Strategy | [references/git-flow.md] |
| Worktrees for Sub-Agents | [references/git-worktree.md] |
| Secrets & Env Config | [references/secrets-config.md] |
| Test Strategy & Quality Gates | [references/test-strategy.md] |
| Changelog & Release Notes | [references/changelog-release-notes.md] |
| Notifications & Alerting | [references/notifications.md] |
| Policy Drift Detection | [references/policy-drift.md] |
| Semantic Versioning | [references/semver.md] |
| ⚡ When to commit / open / close / merge | [agents/decision-engine.md] ← read when unsure |
| Sub-Agent Orchestration | [agents/orchestrator.md] |
| Long Task Management | [agents/long-tasks.md] |
| Conflict Resolution | [agents/conflict-resolver.md] ← when PR hits conflicts |
| Rollback & Recovery | [agents/rollback-playbook.md] ← when things go wrong |
| Agent Health Monitor | [agents/health-monitor.md] ← watchdog for sub-agents |
| Feature Dependency Ordering | [agents/dependency-ordering.md] ← chained features |
| PR Review Automation | [agents/pr-review-automation.md] |
| Agent Audit Log | [agents/audit-log.md] |
| Multi-Repo Atomic Changes | [agents/multi-repo-saga.md] |
| Sprint Kickoff Automation | [agents/sprint-planner.md] ← full sprint in one command |
| Team Leader Agent | [agents/team-leader/SKILL.md] ← manages sub-agents + inter-agent comms |
Authentication
Always establish auth before any ADO operation.
Personal Access Token (PAT)
# Set env vars — never hardcode
export ADO_ORG="https://dev.azure.com/YOUR_ORG"
export ADO_PAT="your-pat-token"
export ADO_PROJECT="YourProject"
# Encode for basic auth header
B64_PAT=$(echo -n ":$ADO_PAT" | base64)
az CLI (preferred for most tasks)
# Install & login
az extension add --name azure-devops
az devops configure --defaults organization=$ADO_ORG project=$ADO_PROJECT
az devops login --organization $ADO_ORG # paste PAT when prompted
Git credential setup for ADO
# Preferred: use the Git Credential Manager (GCM) — no PAT in URLs
git config --global credential.helper manager
az repos clone --repository REPO # az CLI handles auth automatically
# For CI/CD automation without GCM, pass PAT via Git's askpass mechanism
# rather than embedding it in the URL (URLs appear in logs and .git/config)
export GIT_ASKPASS=/usr/lib/git-core/git-askpass
git -c credential.helper= \
-c "url.https://anything:[email protected]=https://dev.azure.com" \
clone https://dev.azure.com/ORG/PROJECT/_git/REPO
# ⚠️ NEVER embed PAT directly in clone URLs — they are stored in .git/config,
# shell history, and CI/CD logs.
Core Principles
- Idempotency — All scripts must be safe to re-run. Check existence before creating.
- Atomicity — For multi-step tasks, checkpoint state so sub-agents can resume.
- Least privilege — Scope PATs to minimum required permissions.
- Audit trail — Log all write operations with timestamps.
- Rate limits — ADO REST API: 200 req/min per user. Add
sleep 0.3between bulk calls.
Workflow Decision Tree
User Request
│
├─ Unsure WHEN to commit, open feature, close worktree, or merge?
│ └─ agents/decision-engine.md ← START HERE
│
├─ PR hit a conflict / rebase failed?
│ └─ agents/conflict-resolver.md ← triage + recovery steps
│
├─ Something went wrong — need to undo / revert / recover?
│ └─ agents/rollback-playbook.md ← branch/PR/release/deploy rollback
│
├─ Sub-agent crashed, stalled, or hung?
│ └─ agents/health-monitor.md ← watchdog + restart logic
│
├─ Feature B depends on Feature A (not merged yet)?
│ └─ agents/dependency-ordering.md ← chained branches + batch order
│
├─ Single repo/branch/PR operation?
│ └─ references/git-ops.md
│
├─ Need a branching strategy / release workflow?
│ └─ references/git-flow.md
│
├─ Sub-agents need isolated git workspaces?
│ └─ references/git-worktree.md
│
├─ Secrets, variable groups, Key Vault, env config?
│ └─ references/secrets-config.md
│
├─ Test coverage / quality gates / static analysis?
│ └─ references/test-strategy.md
│
├─ Generate CHANGELOG or release notes?
│ └─ references/changelog-release-notes.md
│
├─ Send Teams/email alerts for PR/pipeline/deploy events?
│ └─ references/notifications.md
│
├─ Branch policies out of sync across repos?
│ └─ references/policy-drift.md
│
├─ Auto-calculate next version from commits?
│ └─ references/semver.md
│
├─ Auto-assign reviewers, label PR by size, post diff summary?
│ └─ agents/pr-review-automation.md
│
├─ Audit trail of all agent actions?
│ └─ agents/audit-log.md
│
├─ Change must land in 3+ repos simultaneously?
│ └─ agents/multi-repo-saga.md ← saga with rollback
│
├─ Kick off a full sprint from ADO backlog?
│ └─ agents/sprint-planner.md ← tickets → worktrees → agents in one command
│
├─ Need a Team Leader to manage agents + route inter-agent requests + enhance prompts?
│ └─ agents/team-leader/SKILL.md ← start here, then its sub-files
│
├─ Pipeline create/run/monitor?
│ └─ references/pipelines.md
│
├─ Work items / sprints / boards?
│ └─ references/work-items.md
│
├─ Bulk operation (many repos, branches, PRs)?
│ └─ agents/orchestrator.md
│
└─ Long-running task (>2 min, many steps)?
└─ agents/long-tasks.md
Common Quick Commands
# List all repos in project
az repos list --output table
# Create branch
az repos ref create --name "refs/heads/feature/my-branch" \
--object-id $(az repos ref list --filter heads/main --query "[0].objectId" -o tsv) \
--repository MY_REPO
# List open PRs
az repos pr list --status active --output table
# Trigger a pipeline
az pipelines run --name "MyPipeline" --branch main
# List work items in current sprint
az boards query --wiql "SELECT [Id],[Title],[State] FROM WorkItems WHERE [System.IterationPath] = @CurrentIteration AND [System.TeamProject] = '$ADO_PROJECT'"
Reference Files
Load these on demand — only read what's needed for the task:
- references/git-ops.md — clone, push, merge, cherry-pick, rebase, conflict resolution, history rewrite
- references/repos-branches.md — repo creation, branch policies, protection rules, naming conventions
- references/pull-requests.md — PR creation, reviewers, auto-complete, merge strategies, PR templates
- references/pipelines.md — YAML pipeline authoring, stages/jobs/steps, triggers, variables, secrets, agents
- references/work-items.md — create/update/link work items, sprints, queries, bulk import
- references/rest-api.md — raw REST patterns, pagination, retry logic, batch endpoints
- references/git-flow.md — feature/release/hotfix branching model, ADO policy setup, automation scripts
- references/git-worktree.md — parallel isolated workspaces per sub-agent, bare clone setup, full orchestration
- references/secrets-config.md — PAT, variable groups, Key Vault linking, service connections, secret scanning, rotation
- references/test-strategy.md — test pyramid, coverage thresholds by branch, quality gates, agent test runner
- references/changelog-release-notes.md — auto-generate CHANGELOG.md and release notes from conventional commits
- references/notifications.md — Teams webhooks, ADO service hooks, stale PR scanner, pre-built alert templates
- references/policy-drift.md — baseline definition, scanner, auto-remediation, scheduled pipeline
- references/semver.md — compute next version from commits, prerelease versions, pipeline integration
Agent Files
- agents/decision-engine.md — when to commit, open/close features, create/destroy worktrees, merge strategy
- agents/conflict-resolver.md — rebase conflicts, PR conflicts, parallel agent conflicts, escalation
- agents/rollback-playbook.md — branch/PR/release/deployment/ADO-data rollback with decision matrix
- agents/health-monitor.md — heartbeat protocol, watchdog loop, restart logic, health dashboard
- agents/dependency-ordering.md — topological sort, chained branches, contract stubs, batch execution
- agents/pr-review-automation.md — CODEOWNERS reviewer assignment, size labelling, diff summary comment
- agents/audit-log.md — structured event log, AuditLog class, query tool, compliance CSV export
- agents/multi-repo-saga.md — prepare/validate/commit/rollback saga for atomic multi-repo changes
- agents/team-leader/SKILL.md — Team Leader entry point: architecture, role map, quick decision guide
- agents/team-leader/prompt-enhancer.md — rewrite vague tasks into precise instructions before agents start; inject prior agent knowledge; set quality bar; anticipate blockers
- agents/team-leader/message-broker.md — inter-agent request/response protocol; agents never talk directly; TL routes knowledge_request, file_request, clarification_request, blocker_report, review_request, split_request, human_request
- agents/team-leader/knowledge-store.md — shared memory of what every agent produced; auto-ingested from RESULT.json;
find_owner(),get_context_for_agent(), conflict detection - agents/team-leader/task-manager.md — dynamic reassignment, task splitting, duplicate merging, escalation; full TL main event loop
- agents/orchestrator.md — central coordination hub: integrates health-monitor, audit-log, dependency-ordering, canonical sub-agent prompt template, full filesystem layout
- agents/long-tasks.md — checkpoint/resume pattern for tasks that exceed context or time limits
Gives 0 of the 12 instructions most pr commit review skills give
Counted across 888 of the 1,342 authors here whose files we hold, read 2026-08-06
- use conventional commits formatin 123 of 888, across 110 files
- keep subject line under 72 charactersin 60 of 888, across 46 files
- delete branches after mergein 50 of 888, across 37 files
- use imperative mood in subject linein 50 of 888, across 41 files
- use imperative mood in commit messagesin 45 of 888
- generate a conventional commit messagein 42 of 888
- make atomic commitsin 37 of 888, across 25 files
- run tests before committingin 36 of 888, across 24 files
- run project test suite to verify clean baselinein 35 of 888, across 7 files
- run detected project setup commandsin 34 of 888, across 6 files
- wrap commit body at 72 charactersin 32 of 888, across 25 files
- split unrelated changes into separate commitsin 32 of 888, across 27 files
Said here and by no other author read
- Establish authentication before any operation
- Scope personal access tokens to minimum permissions
- Log all write operations with timestamps
- Add sleep between bulk REST API calls
- Make all scripts idempotent and safe to re-run
- Checkpoint state for multi-step tasks
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.