Dev & Engineering github-issuesissue-triageissue-templatesprivacy-sanitizationread-only-discovery

GitHub Issue Creation & Triage

Create, review, and triage GitHub issues based on the repository's actual contribution workflow, avoiding duplicates and privacy leaks.

FollowSkills review · FSRS-2.0
Use with care
60/ 100 5-point scale 3.0 / 5
1 2 3 4 5 6
1Trust20 / 25 · 4.0/5

The skill follows least privilege and user confirmation principles: it requires read-only discovery first (gh auth status, gh repo view, gh api), stops and asks on failure, enforces pre-submission privacy review replacing sensitive data with placeholders, does not invent labels, and has no malicious or overreaching operations. Deductions: external dependency (gh CLI) security not verified, explicit user confirmation prompt step not clearly specified (internal review instead), rollback/recovery not addressed.

2Reliability10 / 20 · 2.5/5

The skill steps are self-consistent, provide a clear discovery process and failure handling (stop and ask), but error handling is limited (no specific error message templates), dependency on external tool (gh) availability unverified; static assessment cannot execute, but instructions seem reproducible. Deductions: no test evidence, failure feedback only says 'stop and ask' without diagnostic details.

3Adaptability11 / 15 · 3.7/5

Use cases clear (creating and triaging GitHub issues), trigger conditions clear, but non-fit scenarios not explicitly stated (e.g., non-GitHub repos). Chinese environment: uses /gh/ commands and English templates, GitHub accessibility from mainland China may be limited, but the skill itself only operates on GitHub repos, may be affected by network. Deductions: no environment adaptation notes or Chinese support.

4Convention9 / 15 · 3.0/5

Documentation structure clear with When To Use, Core Rule, Workflow, but lacks install/dependency notes (assumes gh installed), version info present in front matter (v1.2), but maintenance responsibility and update path not explicit. Deductions: lacks detailed troubleshooting and FAQ, no long-term maintenance plan.

5Effectiveness7 / 15 · 2.3/5

Core task (creating and triaging issues) described in detail with clear step logic, but static assessment cannot verify actual output usability; no example outputs or marginal value evidence. Deductions: no run evidence, output quality and comparative benefit unverified.

6Verifiability3 / 10 · 1.5/5

Only skill documentation and repo-level CI files present, no direct tests or execution records for this skill; key claims (e.g., discovery process effectiveness) based on documentation, lacking independent verification. Deductions: no direct evidence to verify skill behavior.

Evidence confidence:Low Reviewed Aug 07, 2026 Reviewed revision 1eb4b2602105
The upstream repository has new commits since this review. The score still applies to the reviewed revision shown and may not cover the latest changes.
Before you use it
  • The skill depends on GitHub CLI (gh); ensure it is installed and authenticated in the user's environment, otherwise the workflow fails.
  • Publishing external issues has external effects; strictly follow the privacy review to avoid leaking sensitive data.
  • In mainland China network conditions, GitHub access may be unstable; assess network reachability.
Review evidence [1][2][3][4][5][6][7]
See the full review method →

What does this skill do, and when should you use it?

This skill guides an AI agent to discover a repository's contribution conventions (templates, labels, approval gates, Discussions support) via read-only commands before proposing or publishing an issue. It requires checking CONTRIBUTING.md, README, issue templates, and label lists to ensure compliance with repository policy, and stops to ask when unsure. The skill emphasizes not inventing info, labels, or processes, and enforces a pre-submission privacy review that replaces environment-specific data with placeholders. It is one of 38 skills in the gentle-ai monorepo, which is released under the MIT license, while this specific skill is Apache-2.0.

Runs read-only GitHub CLI commands (gh auth status, gh repo view) to verify auth, resolve repo info, and check for Discussions/Issues enablement and label lists. Reads repository files like CONTRIBUTING.md, README, .github/ISSUE_TEMPLATE, and config.yml. Searches open and closed issues for duplicates. Builds label arguments from discovered labels using --label. Performs a privacy review before publishing, replacing private project names, usernames, hostnames, paths, tokens, etc. with placeholders. Publishes via gh issue create, optionally with --web for forms or --body-file for Markdown templates. Does not publish without a template unless blank issues are explicitly enabled.

  1. A developer encountering a bug in a public repo with strict contribution rules wants to file a compliant issue without getting rejected.
  2. An AI agent working on an existing open-source project needs to propose a feature request without duplicating an existing one.
  3. A team member in an internal repo needs to report a problem but wants to avoid leaking internal project names, usernames, or internal hostnames.
  4. A maintainer triaging new issues needs a checklist to verify an issue is concrete, has evidence, and is in scope.
  5. A user unsure whether to use an issue template or ask a question wants the skill to decide if it should go to Discussions or contact links.

What are this skill's strengths and limitations?

Pros
  • Enforces a discovery-first approach to avoid violating repository policies.
  • Prevents duplicate issues by searching existing ones.
  • Built-in privacy review that identifies and replaces environment-specific data.
  • Handles no-template scenarios clearly, never publishing without permission.
  • Read-only operations for safety.
Limitations
  • Depends entirely on GitHub CLI (gh), which must be installed and authenticated.
  • Stops and asks when policy discovery fails, potentially increasing interaction overhead.
  • Privacy review requires user confirmation for correct replacement.
  • No evidence of an automated test suite, potentially lacking verification.
  • Does not mention support for GitHub Enterprise or non-GitHub platforms.

How do you install this skill?

This skill is provided as part of the gentle-ai repository, which bundles 38 skills. Install the entire collection via the official installer (macOS/Linux: curl -fsSL https://raw.githubusercontent.com/Gentleman-Programming/gentle-ai/main/scripts/install.sh | bash; Windows PowerShell: irm ... | iex). After installation, copy this skill's directory (internal/assets/skills/issue-creation/) into your Agent Skills folder, or run gentle-ai skill-registry refresh to update the registry.

How do you use this skill?

In any GitHub repository, prompt the agent with a trigger like 'create an issue' or 'file a bug report'. The skill will run discovery commands, check contribution guidelines, and if blank issues are not enabled or templates are missing, it will ask for context or follow the repo's contact links, rather than guessing. Follow the skill's instructions, provide necessary info (problem description, reproduction steps, environment), and confirm the privacy replacements it generates. Before publishing, the skill runs privacy review and creates the issue.

FAQ

What permissions does this skill require?
It requires GitHub CLI (gh) installed and authenticated locally with access to the target repo. The skill performs read-only operations (viewing repo info, listing labels) and a final creation command. It does not require additional permissions, but needs network access to GitHub API.
What if the repository has issues disabled?
The skill detects hasIssuesEnabled as false and stops, asking you for the repo's contact links or how to report issues. It does not attempt to publish an issue.
Does the privacy review remove all sensitive information?
Yes, it replaces private project names, usernames, hostnames, paths, and tokens with placeholders like <project-name> and <user>. However, it does not remove intentionally public identifiers like tool names or public documentation URLs. If a replacement makes a reproduction step impossible, it marks it as <value-required> and asks you to fill it in.
Is this skill portable to other platforms?
The skill is designed to be platform-agnostic, relying on GitHub CLI rather than any specific agent platform. Therefore, it should work on any agent that can run shell commands.

More skills from this repository

All from Gentleman-Programming/gentle-ai

Related skills