home / skills / serejaris / ris-claude-code / gh-issues

gh-issues skill

/skills/gh-issues

This skill helps you manage GitHub issues via gh CLI, track AI session context, and streamline task workflows across labels.

npx playbooks add skill serejaris/ris-claude-code --skill gh-issues

Review the files below or copy the command above to add this skill to your agents.

Files (4)
SKILL.md
7.1 KB
---
name: gh-issues
description: >-
  Use when creating, searching, updating, or managing GitHub issues via CLI.
  Triggers: "issue", "create issue", "gh issue", "task tracking",
  "context", "handoff", "resume task", "session context", "save progress",
  "active tasks", "in-progress", "my tasks", "open issues".
  Covers: gh commands, bulk operations, JSON/jq, search filters, issue-to-PR workflow,
  AI session context storage, task workflow with labels.
---

# GitHub Issues CLI

Efficient GitHub Issues management via `gh` CLI with AI session context storage.

## Table of Contents

1. [Quick Reference](#quick-reference)
2. [JSON Output](#json-output-patterns)
3. [Search Filters](#advanced-search-filters)
4. [Bulk Operations](#bulk-operations)
5. [Issue to PR Workflow](#issue-to-pr-workflow)
6. [Milestones](#milestones-via-api)
7. [AI Session Context](#ai-session-context)
8. [Task Workflow](#task-workflow)
9. [Examples](#examples)
10. [Common Mistakes](#common-mistakes)

## Quick Reference

| Task | Command |
|------|---------|
| Create issue | `gh issue create -t "Title" -b "Body" -l bug -a @me` |
| **Active tasks** | `gh issue list -l in-progress -s open` |
| List open bugs | `gh issue list -l bug -s open` |
| View as JSON | `gh issue view 123 --json number,title,body,labels,state` |
| Close with comment | `gh issue close 123 -c "Fixed in #456"` |
| Edit labels | `gh issue edit 123 --add-label priority:high` |
| Start work | `gh issue edit 45 --add-label in-progress` |
| Create branch | `gh issue develop 123 --checkout` |
| Load context | `gh issue view 45 --json comments --jq '.comments[] \| select(.body \| contains("AI-CONTEXT"))' ` |

## JSON Output Patterns

Always use `--json` for parsing. Fields: `number`, `title`, `body`, `state`, `labels`, `assignees`, `milestone`, `author`, `createdAt`, `updatedAt`, `comments`, `url`.

```bash
gh issue view 123 --json number,title,labels,state
gh issue list --json number,title,labels --jq '.[] | select(.labels[].name == "bug")'
gh issue list -l bug --json number --jq '.[].number'
```

## Advanced Search Filters

```bash
gh issue list --search "is:open author:username"
gh issue list --search "created:>=2026-01-01 created:<=2026-01-07"
gh issue list --search "label:bug label:priority:high"
gh issue list --search "is:open -label:wontfix"
gh issue list --search "milestone:v2.0"
```

## Bulk Operations

```bash
gh issue edit 10 15 20 --add-label "priority:high"
gh issue close 10 15 20 -c "Duplicate of #5"
gh issue list -l needs-triage --json number --jq '.[].number' | \
  xargs -I{} gh issue edit {} --add-label reviewed
```

## Issue to PR Workflow

```bash
gh issue develop 123 --checkout          # Create branch from issue
git add . && git commit -m "fix: #123"   # Commit with reference
gh pr create --fill                       # Create PR (auto-links)
gh issue close 123 -c "Fixed in PR #456" # Close when merged
```

## Milestones via API

```bash
gh api repos/:owner/:repo/milestones --jq '.[].title'
gh issue edit 123 -m "v2.0"
```

## AI Session Context

Store session context in GitHub issues for seamless task handoff.

### Read Context

```bash
# Get AI context from issue comments
gh issue view 45 --json comments --jq '
  .comments[] | select(.body | contains("AI-CONTEXT:START")) | .body
'

# Get comment ID for updates
gh issue view 45 --json comments --jq '
  .comments[] | select(.body | contains("AI-CONTEXT:START")) | .id
'
```

### Save Context

```bash
# Create new context comment
gh issue comment 45 --body-file .ai-context.md

# Update existing (replace COMMENT_ID)
gh api repos/:owner/:repo/issues/comments/COMMENT_ID \
  --method PATCH -f body="$(cat .ai-context.md)"
```

### Context Template

See `references/context-template.md` for full template. Minimal version:

```markdown
<!-- AI-CONTEXT:START -->
## Context | IN_PROGRESS
**Files:** `file.py:45`, `other.py:120`
**Done:** task1, task2
**Next:** next task
**Resume:** One-line summary for cold start
<!-- AI-CONTEXT:END -->
```

### Workflow

1. **Start work** — load context from issue
2. **Work on task** — track progress mentally
3. **Pause/Stop** — save context to issue comment
4. **Resume later** — load context, continue

## Task Workflow

Manage issue lifecycle with labels.

### Labels

| Label | Meaning |
|-------|---------|
| `backlog` | In queue |
| `in-progress` | Active work |
| `blocked` | Blocked |
| `review` | Needs review |

### View Active Tasks

```bash
# My active issues
gh issue list -l in-progress -s open

# All my assigned issues
gh issue list --assignee @me -s open

# Issues with context
gh issue list -s open --json number,title,labels --jq '
  .[] | select(.labels[].name == "in-progress") | "#\(.number): \(.title)"
'
```

### Start Working

```bash
# Mark as in-progress
gh issue edit 45 --add-label in-progress --remove-label backlog

# View issue + load context
gh issue view 45 --json number,title,body,comments --jq '{
  number, title, body,
  context: (.comments[] | select(.body | contains("AI-CONTEXT")) | .body) // "No context"
}'
```

### Pause/Block

```bash
# Mark as blocked
gh issue edit 45 --add-label blocked --remove-label in-progress

# Add blocking comment
gh issue comment 45 --body "Blocked: waiting for API access"
```

### Complete

```bash
# Close with comment
gh issue close 45 -c "Done in commit abc123"

# Or close via PR (auto-closes if PR body contains "Fixes #45")
```

## Examples

### Create Issue for Task

```bash
gh issue create -t "Simplify bot flow" -b "## Problem
Bot stopped converting. Current flow is complex.

## Reference
Simple flow example from competitor.

## Tasks
- [ ] Analyze current flow
- [ ] Create new diagram
- [ ] Implement"
```

### Start Working on Issue

```bash
# 1. View issue with context
gh issue view 45 --json number,title,body,comments --jq '{
  number, title, body,
  context: (.comments[] | select(.body | contains("AI-CONTEXT")) | .body) // "No context"
}'

# 2. Create branch
gh issue develop 45 --checkout

# 3. Work...
```

### Save Progress Before Stopping

```bash
# Create context file
cat > .ai-context.md << 'EOF'
<!-- AI-CONTEXT:START -->
## AI Session Context
_Last updated: 2026-01-07 16:00_

### Status
**IN_PROGRESS**

### Progress
- [x] Analyzed current flow
- [ ] Create new diagram

### Key Files
- `handlers/start.py:45-78` — start handler
- `data/courses.yaml` — config

### Next Steps
1. Draw simplified flow
2. Implement changes

### Resume Context
Simplifying bot flow. Analyzed current state. Next: create diagram.
<!-- AI-CONTEXT:END -->
EOF

# Save to issue
gh issue comment 45 --body-file .ai-context.md
```

### Resume Work

```bash
# Load context
CONTEXT=$(gh issue view 45 --json comments --jq '
  .comments[] | select(.body | contains("AI-CONTEXT")) | .body
' 2>/dev/null)

echo "$CONTEXT"
# Continue from where you left off...
```

## Common Mistakes

| Mistake | Fix |
|---------|-----|
| Missing `--json` | Always `--json field1,field2` |
| Loop instead of multi-arg | `gh issue edit 1 2 3` |
| Using `grep` on output | Use `--jq` |
| Forgot `-s all` for closed | Default is open only |
| `{owner}/{repo}` in API | Use `:owner/:repo` |
| Duplicate context comments | Update existing, don't create new |

Overview

This skill provides concise, practical guidance for managing GitHub Issues via the gh CLI, including creating, searching, bulk-editing, and integrating issue-based AI session context. It streamlines task workflows with label conventions, issue-to-PR patterns, and JSON/jq parsing for reliable automation. Use it to keep task state, hand off work, and resume progress with minimal friction.

How this skill works

The skill shows gh command patterns to create, view, edit, comment, and close issues, with examples that favor --json and --jq for machine-parsable output. It documents bulk operations (multi-arg edit/close and xargs pipelines), advanced search filters, and an issue-to-PR workflow that links commits and PRs back to issues. It also defines a simple AI-CONTEXT comment template and commands to read, create, and update context stored in issue comments.

When to use it

  • Creating or updating issues from the CLI with labels, assignees, and body content
  • Searching issues with date, label, assignee, or milestone filters and extracting IDs with --json/--jq
  • Performing bulk edits or bulk-closing multiple issues in a single command
  • Saving and loading AI session context into issue comments for handoff and resume
  • Starting work on an issue, creating a branch tied to the issue, and following the issue-to-PR flow

Best practices

  • Always use --json and --jq for parsing output to avoid brittle text scraping
  • Use consistent labels (backlog, in-progress, blocked, review) to represent lifecycle state
  • Update an existing AI-CONTEXT comment instead of appending duplicates to keep context authoritative
  • Reference issue numbers in commits (e.g., commit message: fix: #123) so PRs auto-link and close issues
  • Use multi-arg gh commands (gh issue edit 1 2 3) for efficient bulk operations instead of loops

Example use cases

  • Start work: load AI context from an issue comment, add in-progress label, create branch with gh issue develop
  • Pause and save work: write a .ai-context.md file and gh issue comment --body-file to persist session state
  • Bulk tag triaged issues: gh issue list -l needs-triage --json number | jq -r '.[].number' | xargs gh issue edit --add-label reviewed
  • Close issues from a release: gh issue close 10 15 20 -c 'Fixed in release v2.0'
  • Search for recent activity: gh issue list --search "created:>=2026-01-01 created:<=2026-01-07" --json number,title

FAQ

How do I avoid duplicate AI-CONTEXT comments?

Query comments for the AI-CONTEXT marker to get the comment ID, then use gh api PATCH to update that comment instead of creating a new one.

What fields should I request with --json?

Request number,title,body,labels,state,assignees,comments,createdAt,updatedAt and any other fields you need for automation.