github-manager skill
✓ VerifiedConfigures and optimizes GitHub repositories for cost-effective CI/CD, issue tracking, branch protection, and release workflows.
- GitHub stars
- 13
- Files
- 1
- Last commit
- Nov 9, 2026
- License
- Apache-2.0
- Allowed tools
- —
What does the github-manager skill do?
github-manager configures and optimizes GitHub repositories for cost-effective CI/CD, issue tracking, branch protection, and release workflows.
Summary generated from the skill's SKILL.md.
About the github-manager skill
Managing GitHub repository settings to control costs and enforce delivery standards is what this skill enables. It helps developers and DevOps engineers move away from unmanaged automation toward a governed environment where CI triggers, branch protection, and issue tracking follow strict rules. An agent should load this skill when it needs to audit GitHub Actions spending, design a label taxonomy, or configure repository-level enforcement mechanisms like required status checks.
The skill produces concrete configuration outputs such as CI trigger matrices, branch-protection policies, and release checklists. It focuses on optimizing GitHub Actions billing by accounting for runner multipliers, specifically noting that macOS jobs cost ten times more than Linux. By implementing path filters and selective triggers for pull requests and version tags, it prevents unnecessary spending on feature branches. It also establishes structured issue tracking through namespaced labels and mandatory acceptance criteria to ensure traceable development workflows.
What an agent can do with it
- 1Standardize issue organizationCreate a consistent label taxonomy to prevent noise in project tracking.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the github-manager skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the github-manager skill in Plugin
/plugin marketplace add jpantsjoha/ai-native-developer-experience /plugin install devex-charter@join-the-team-marketplace
Install the github-manager skill in Claude Code
~/.claude/skills/github-manager/SKILL.md
$ npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager ~/.claude/skills/github-managerInstall the github-manager skill in Claude Code (project)
.claude/skills/github-manager/SKILL.md
$ npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager .claude/skills/github-managerInstall the github-manager skill in Codex
~/.agents/skills/github-manager/SKILL.md
$ npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager ~/.agents/skills/github-managerInstall the github-manager skill in Gemini CLI
~/.gemini/skills/github-manager/SKILL.md
$ npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager ~/.gemini/skills/github-managerInstall the github-manager skill in Claude.ai
npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager github-manager && zip -r github-manager.zip github-manager # claude.ai → Customize → Skills → + → Create skill → Upload a skill → github-manager.zip
Automated checks
Ran Nov 16, 2026Full report →Passed our automated checks. That is not a security audit: read the scripts before you let an agent run them. How checks work
- Source repository linked and reachablejpantsjoha/ai-native-developer-experience/skills/github-manager is public and reachablegithub.com/jpantsjoha/ai-native-developer-experience/tree/HE ↗Pass
- Open source license (OSI)Apache-2.0spdx.org/licenses/Apache-2.0.html ↗Pass
- Commit or release in the last yearLast commit 2026-11-09github.com/jpantsjoha/ai-native-developer-experience/commits ↗Pass
- No download-and-run or credential theft in scriptsNo scripts in the skill folderPass
- No hidden instructions in SKILL.mdNo override phrases, safeguard switches, credential requests or hidden charactersgithub.com/jpantsjoha/ai-native-developer-experience/blob/HE ↗Pass
- Pre-approved tools are scopedNo pre-approved tools; the agent asks before using onegithub.com/jpantsjoha/ai-native-developer-experience/blob/HE ↗Pass
- SKILL.md follows the Agent Skills formatValid name and description; says when to use it; 124 linesgithub.com/jpantsjoha/ai-native-developer-experience/blob/HE ↗Pass
SKILL.md
View on GitHubGitHub Manager
CI validates what cannot be cheaply verified locally. It is not a remote test runner for every commit.
This skill enforces disciplined, cost-aware GitHub operations. It turns GitHub from a free-for-all into a governed delivery surface: selective automation, traceable issues, consistent labels, and protected mainlines.
When to use
- Setting up or revising a repository's CI/CD workflow
- Investigating unexpected GitHub Actions spend
- Creating or triaging issues and defining acceptance criteria
- Designing a label taxonomy or branch-protection policy
- Preparing a release process or tag strategy
Operating model context
This skill governs the GitHub surfaces that enforce the harness contract at the repository level. It is not general DevOps housekeeping — each procedure below maps to a harness invariant:
- Branch protection + required status checks are the enforcement mechanism for the PR review gate and the exact-candidate binding rule. A CI check that passes on an unprotected branch is a claim; a passing check required by branch rules is evidence.
- Receipts are delivery evidence. Both CI and an explicitly approved controlled runner must retain candidate/artifact-bound commands, results and provenance. A green label or local success claim alone does not establish release readiness.
- Modifying branch protection, CI pipelines, or billing settings is an R2 action. These changes affect all contributors and shared infrastructure. Classify risk, confirm authority, and record the decision before any write.
Use release-readiness to gate a specific deployment. Use this skill to configure and
audit the repository surfaces that make those gates trustworthy.
Procedure
1. Rightsize CI triggers
Resolve the approved release ADR and profile first. The following CI layout is the default recommendation; audit and configure the mode the team actually approved. Do not silently reinstate retired CI, substitute a runner during an outage, or treat local execution as equivalent without review of the required evidence.
- Push to default branches should run lightweight gates only (lint, typecheck, unit tests).
- Pull requests and version tags run the full pipeline, including packaging and cross-platform smoke tests.
- Feature branches do not fire CI on every push; open a PR when ready for validation.
- Path filters skip irrelevant jobs: docs-only changes should not rebuild the extension, code-only changes should not re-render documentation.
2. Minimise Actions billing
- Know the runner multipliers: Linux ×1, Windows ×2, macOS ×10.
- Gate expensive runners (macOS, Windows) behind PRs and releases, not every push.
- Cache dependency stores (
cache: 'pnpm',cache: 'npm', pip, etc.). - Do not run local-integration or exploratory tests in CI; keep them local or on-demand.
- Set a spending limit and billing alert before usage surprises you.
3. Structure issues and tracking
- Every issue states a problem or outcome and has acceptance criteria.
- Link PRs to issues (
Closes #123orRefs #123). - Close issues with a concise note explaining what changed and where.
- Use milestones or projects for release-scope tracking, not long-lived catch-all issues.
4. Label consistently
Adopt a namespaced taxonomy and avoid one-off labels:
| Namespace | Examples | Purpose |
|---|---|---|
kind/ |
kind/bug, kind/feature, kind/docs |
Type of work |
area/ |
area/ci, area/ui, area/security |
Component or domain |
priority/ |
priority/p0, priority/p1 |
Triage urgency |
status/ |
status/blocked, status/needs-review |
Workflow state |
5. Protect the mainline
- Require PR reviews before merging to
main. - Require status checks to pass (lint, typecheck, test).
- Use squash or rebase merges for a linear history; avoid merge commits unless the project explicitly allows them.
- Restrict force-push and deletion on default branches.
6. Define the release workflow
- In
tag-cimode, version tags trigger CI builds. Promotion needs its own evidence and release-owner approval boundary, even if the same tag starts that pipeline. - In
approved-alternativemode, follow the ADR's trigger and controlled executor. Validate the same candidate/artifact, review, rollback and observation requirements. - Reconcile the effective pipeline with the profile; invoke
release-managerand its structural contract check. - Verify version/changelog alignment and protect tags with tag rules, not merely branch protection. Neither a label nor a successful build authorises deployment.
Outputs
- CI trigger matrix: event → jobs that run
- Monthly Actions cost estimate with runner-multiplier breakdown
- Issue template with acceptance criteria
- Label taxonomy
- Branch-protection policy
- Release checklist
Guardrails
- Local testing is not CI. If a check belongs on a developer's machine, do not run it on every push.
- macOS minutes are the silent budget killer. A one-minute macOS job costs ten billable minutes.
- Unconditional full pipelines scale badly. Every unconditional job is a tax on every future commit.
- Labels without a taxonomy become noise. Delete or consolidate labels that do not fit the scheme.
- Never substitute an unapproved build path. Reproducibility and provenance must be demonstrated for the declared executor, whether hosted or locally controlled.
Anti-rationalization table
| Excuse | Counter |
|---|---|
| "Run everything on every push to be safe" | Safety is selective gates, not redundant burn. Run full checks on PRs and tags. |
| "It's only a few CI minutes" | At 10× for macOS, "a few minutes" becomes hundreds of dollars per cycle. |
| "We'll clean up labels later" | Label debt compounds fast and breaks automation that depends on them. |
| "Force-push is fine, we're a small team" | Force-push on main destroys recovery options. Protect the branch. |
| "I'll build the release artifact locally" | The approved mode and evidence requirements decide whether that executor is allowed. |
Files in this skill
1github-manager skill questions
How do I install the github-manager skill in Claude Code?
Run npx degit jpantsjoha/ai-native-developer-experience/skills/github-manager ~/.claude/skills/github-manager. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the github-manager skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
How can I reduce GitHub Actions spending?
You can minimize costs by gating expensive runners like macOS and Windows behind PRs and releases, using dependency caching, and setting spending limits and billing alerts.
What is the recommended CI trigger strategy?
Default branches should run lightweight gates like linting and unit tests, while full pipelines including packaging and smoke tests should be reserved for pull requests and version tags.
How should labels be organized in a repository?
Labels should follow a namespaced taxonomy to avoid noise, using prefixes like kind/, area/, priority/, and status/.