- Home
- Skills
- Shakes Tzd
- Contextune
- Copilot Delegate
copilot-delegate_skill
- HTML
4
GitHub Stars
8
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill shakes-tzd/contextune --skill copilot-delegate- FINAL_RECOMMENDATION.md10.9 KB
- INTEGRATION_SUMMARY.md15.1 KB
- OLD_VS_NEW_DESIGN.md10.7 KB
- PARALLEL_INTEGRATION_ANALYSIS.md26.4 KB
- README.md11.1 KB
- REVISED_ARCHITECTURE.md13.9 KB
- SIMPLIFIED_WORKFLOW.md16.3 KB
- SKILL.md10.6 KB
Overview
This skill delegates GitHub operations and research tasks to the GitHub Copilot CLI to preserve Claude Code session capacity. It automatically activates when prompts include GitHub or research keywords, runs Copilot in a subprocess, and returns compressed JSON results. Use it to offload bulk data retrieval, repo changes, and web research so the main Claude session stays available for high-value reasoning.
How this skill works
When a user prompt triggers configured keywords (issue, PR, repo, commit, research, compare, best practices, lookup), the skill launches a Copilot CLI subprocess with a concrete task script or task file. Copilot executes gh CLI commands or web research, writes structured JSON and metadata files to copilot-results/, and the skill returns a compressed summary to Claude. This reduces prompt consumption and presents concise, reviewable outputs for follow-up actions.
When to use it
- Create or batch GitHub issues, PRs, branches, or queries
- List or query commit history and repo statistics in bulk
- Compare libraries, lookup documentation, or gather best-practices research
- Offload repetitive or data-heavy tasks when Claude session limits are near
- Preserve Claude for architecture, security reviews, or complex debugging
Best practices
- Batch related operations into a single task file to minimize overhead
- Use specific, constrained prompts (fields/criteria) for research tasks
- Always review Copilot JSON output and metadata before applying changes
- Monitor Claude session capacity and delegate aggressively as limits approach
- Keep critical architecture and context-dependent work in Claude, not delegated
Example use cases
- Create 20 issues via a single task file to report triaged bugs
- Delegate a three-way library comparison (zustand, jotai, recoil) with specified criteria
- Generate a PR from feature branch to main with automated description and issue links
- Query last 50 commits with authors and dates and receive compact JSON for review
- Lookup current documentation for Next.js App Router server components and return summarized findings
FAQ
Typical delegation moves multi-prompt research or bulk GitHub operations into a single Claude prompt, often saving 50–80% of prompt consumption depending on task size.
What format are results returned in?
Delegation scripts write structured JSON result files and a .meta file with status, duration, timestamp, and token estimates; the skill returns a compressed summary for quick review.