Skip to content

awesome-claude-code submission: form values (submit 2026-04-26+) #63

@verkyyi

Description

@verkyyi

Form values drafted 2026-04-21, ready for submission on or after Sunday 2026-04-26 (form requires repo ≥ 1 week old; our first commit was 2026-04-19).

Submission URL (web UI only — the maintainer bans gh CLI submissions and the repo has a strict spam-deterrent system):
https://github.com/hesreallyhim/awesome-claude-code/issues/new?template=recommend-resource.yml


Fields

Display Name

github-agent-runner

Category

Agent Skills

(Maintainer note in the form: "I'm currently lumping most things called 'plugins' under Agent Skills until I figure out a better classification system.")

Sub-Category

General

Primary Link

https://github.com/verkyyi/github-agent-runner

Author Name

verkyyi

(Swap to "Verky Yi" if preferred as the display name.)

Author Link

https://github.com/verkyyi

License

MIT

Description

Conversational discovery and install of GitHub Agentic Workflows (gh-aw) from the githubnext/agentics catalog. Includes an agent-team pattern — four coordinating workflows (spec, plan, implementation, review) that collaborate through comments on a single GitHub issue, installable in one pass. The OAuth auth path for subscription users runs the official Claude CLI inside GitHub Actions, the configuration Anthropic documents and permits.

Validate Claims (mandatory)

Install: add the marketplace (https://raw.githubusercontent.com/verkyyi/github-agent-runner/main/.claude-plugin/marketplace.json) and run /plugin install github-agent-runner. Uninstall: /plugin uninstall github-agent-runner, and optionally gh aw remove <workflow> for any installed workflows. Validation coverage lives in tests/ (11 bash e2e + unit scripts; run-tests.sh is the entry point). The agent-team pipeline is validated end-to-end in the public playground: github.com/verkyyi/agent-team-playground PR #5.

Specific Task(s) (mandatory)

1. Point Claude at a fresh repo you own and ask which agentic workflows would be useful. The /discover-workflows skill inspects the repo (language, test presence, CI config, recent activity) and returns 1–3 catalog recommendations with a one-sentence reason per recommendation.

2. Pick daily-repo-status from the recommendations and ask for it to be installed. The /install-workflow skill walks auth setup (OAuth subscription path or API key), adds the workflow via gh aw add, applies the two-pass OAuth tweak if using the subscription path, and runs gh aw validate.

3. (Heavier, optional) Ask for the full agent-team pattern installed. /install-agent-team adds four workflows, seven labels, and one auth secret atomically. Then open an issue, add the agent-team label, and watch the spec → plan → impl → review pipeline run as four linked workflow runs + a PR.

Specific Prompt(s) (mandatory)

1. "Recommend agentic workflows for this repo." — triggers /discover-workflows.
2. "Install daily-repo-status." — triggers /install-workflow.
3. "Install the full agent-team pattern." — triggers /install-agent-team.

Additional Comments

The plugin is a thin layer over github/gh-aw (GitHub Next's agentic-workflow compiler). It does not replace or wrap the compiler; it adds conversational discovery, an auth decision tree, and the multi-workflow agent-team pattern. Network access from the plugin is limited to GitHub (via gh CLI) and the githubnext/agentics catalog (via raw.githubusercontent.com) — no direct Anthropic API calls from plugin code. Installed workflows, once running in GitHub Actions, call the official Claude CLI.

Checklist (all required, tick each on the form)

  • I have checked that this resource hasn't already been submitted
  • It has been over one week since the first public commit to the repo I am recommending (only true on or after 2026-04-26)
  • All provided links are working and publicly accessible
  • I do NOT have any other open issues in this repository (verify before submitting — close any before opening this one)
  • I am primarily composed of human-y stuff and not electrical circuits

Pre-submission self-eval summary (2026-04-21)

Ran the maintainer's .claude/commands/evaluate-repository.md prompt against the repo. Overall: 8.5/10 → Recommend.

Three pre-submission remedies shipped today:

  • ✅ Uninstall section added to README (commit 7f41566)
  • ✅ Playground demo link added to both README + catalog/agent-team/README.md (commit 7f41566)
  • ⚠️ GIF/screencast deferred — the playground PR link serves the same purpose for this text-output pipeline; record later if desired.

Reminders before submitting

  1. Web UI only. Do not use gh issue create on the awesome-claude-code repo.
  2. Close any open issues you have on that repo before submitting (checklist requires none).
  3. After submission passes the bot validation, wait — don't ping, don't comment asking for status.
  4. If approved, add the "Mentioned in Awesome Claude Code" badge to this repo's README.

Close this issue once the submission is filed.

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentation

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions