Implement marketingskills/open-source-growth agent skills to automate repo audits, README upgrades, demos and ecosystem PRs

TL;DR in plain English

• This guide shows how to add a small "growth skills" toolbox to a repository so an agent can open useful issues and pull requests (PRs). The example implementation and templates live at https://github.com/marketingskills/open-source-growth.
• The toolbox runs a repo audit, upgrades the README, builds a demo, prepares a launch pack, and creates ecosystem PRs (these skill categories are described in the example repo). See the repo for the concrete skill names and templates: https://github.com/marketingskills/open-source-growth.
• Quick start (high level): clone the example repo, copy the example skills configuration, add your API key and GitHub token, run the audit skill and inspect the top suggestions.

Methodology: recommendations below are derived from the example repo and its listed skill categories at the linked repository.

What you will build and why it helps

You will assemble and run a small set of agent "skills" that produce concrete repo artifacts that improve onboarding and discoverability. The example repo documents a set of skills and templates; use it as a starting point: https://github.com/marketingskills/open-source-growth.

Key outputs (as implemented in the example repo) include a prioritized audit, a README PR, a demo directory, a launch pack, and one or more ecosystem-inclusion PRs. A compact comparison table:

| Skill (example) | Primary artifact | Why it helps | |---|---:|---| | repo-audit | audit-report.md | Prioritizes high-impact fixes so maintainers see where to focus | | README upgrade | README PR (diff) | Improves first-run experience and onboarding | | demo builder | demo/ + Dockerfile or Codespaces config | Lets users run the project quickly to evaluate it | | launch-pack | launch-pack/checklist.md | Produces short release notes and an announcement checklist | | ecosystem PRs | registry PRs (1–3 templates) | Increases discoverability in relevant registries |

Each of the above skill categories and templates appears in the example repository; inspect the files and templates there for concrete prompts and scripts: https://github.com/marketingskills/open-source-growth.

Before you start (time, cost, prerequisites)

Prerequisites (from the example repo's approach):

• A public GitHub repository and a GitHub personal access token (PAT). See the repo for templates and expected scopes: https://github.com/marketingskills/open-source-growth.
• An API key for the model you will use; the example config points at a model.name field in skills/skills.yaml in the repo.
• A local clone of the example project for templates and scripts: git clone https://github.com/marketingskills/open-source-growth.git

Quick prep checklist:

• [ ] Clone the repo: git clone https://github.com/marketingskills/open-source-growth.git
• [ ] Create a GitHub PAT and store it as a secret (see the example repo for guidance)
• [ ] Add your model API key to your secret store and reference it in skills/skills.yaml
• [ ] Copy skills/example-skills.yaml → skills/skills.yaml and edit to match your repo

Step-by-step setup and implementation

1. Clone and inspect the toolbox

git clone https://github.com/marketingskills/open-source-growth.git
cd open-source-growth
ls -la

2. Copy the example skills config and edit it. The example config and comments live in the repo: https://github.com/marketingskills/open-source-growth

cp skills/example-skills.yaml skills/skills.yaml
# then edit skills/skills.yaml to set model.name, api_key_secret, github.repo, and pat_secret

Example config fragment (edit to match your secrets and repo):

# skills/skills.yaml (example fragment)
model:
  name: "your-model-name"
  # tune provider-specific settings here
github:
  repo: "your-org/your-repo"
  pat_secret: "GITHUB_PAT"
retry:
  attempts: 3
  backoff_ms: 500

3. Run the audit skill locally to produce a prioritized audit (artifact: audit-report.md). The example repo contains scripts and a runner; follow those files in https://github.com/marketingskills/open-source-growth.

./scripts/run-audit.sh --config skills/skills.yaml --out audit-report.md
# or, if the repo provides a Python runner:
python tools/run_skill.py audit --config skills/skills.yaml --output audit-report.md

4. Review the audit report, run the README upgrade flow, and inspect diffs produced by the README skill. Keep changes focused and small.

5. Run the demo builder skill to generate demo/, a Dockerfile or Codespaces config, and CI entries if available. Validate the demo locally or in Codespaces.

6. Generate the launch pack artifacts and use the checklist to prepare a short announcement.

7. Optionally automate with a GitHub Actions workflow similar to the example workflow in the repo. Example workflow fragment (place in .github/workflows/agent-apply.yml):

name: Agent Apply
on:
  schedule:
    - cron: '0 12 * * 0' # sample schedule (edit in your repo)
jobs:
  apply:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Run agent
        run: scripts/run-agent.sh --config skills/skills.yaml

8. Use a small canary and human-review gate before merging automated PRs. The example repo provides templates and patterns to adapt: https://github.com/marketingskills/open-source-growth.

Common problems and quick fixes

• Noisy or irrelevant PR content

  • Fix: tighten the agent's instructions and examples in skills/skills.yaml; require a human review label before merge.

• Model hallucinations in code snippets

  • Fix: enforce CI unit tests and block merges for failing tests. Keep PRs small and review diffs.

• Rate limits or API errors

  • Fix: add retry/backoff settings in your configuration and instrument retries in the runner. The example repo shows retry patterns: https://github.com/marketingskills/open-source-growth.

• Too many small PRs

  • Fix: set batching rules in the workflow; group related edits and throttle automated PRs.

Quick remediation examples

{"retry": {"attempts": 3, "backoff_ms": 500}}

And a short checklist template for immediate steps:

• [ ] Tighten prompts with explicit examples
• [ ] Add CI checks that block merges
• [ ] Enable a human-approval label for the first N runs
• [ ] Run agent in a canary repo first

Reference templates and examples: https://github.com/marketingskills/open-source-growth

First use case for a small team

Use case: a small open-source team wants faster adoption with minimal ongoing effort. The recommended initial focus (as demonstrated by the example repo) is: run the repo audit, apply the top README suggestions, and validate a demo artifact. See the example repository for templates and skill names: https://github.com/marketingskills/open-source-growth.

Plan (concrete steps you can follow):

1. Run the repo audit skill and review audit-report.md for the highest-priority items.
2. Accept the top README diffs produced by the README skill and open PR reviews for those changes.
3. Merge the demo PR after at least one human review and validate the demo locally or in Codespaces.
4. Run the launch-pack skill to produce release notes and a checklist to prepare an announcement.

Small-team checklist (adapt from the example repo):

• [ ] Run audit and review audit-report.md
• [ ] Accept top README diffs and open PR reviews
• [ ] Merge demo after human review and validate locally
• [ ] Run launch-pack and prepare a short announcement

Track simple impact metrics such as PR conversion rate and demo-run success; iterate weekly using the templates in https://github.com/marketingskills/open-source-growth.

Technical notes (optional)

• Model and prompt tuning: the example repo expects a model.name field and provider-specific parameters in skills/skills.yaml; use the repository to learn where to place your provider settings: https://github.com/marketingskills/open-source-growth.
• Secrets: store API keys and PATs in GitHub Secrets and restrict PAT scopes to the minimum required. The example repository shows how to reference secrets from the skills configuration.
• Observability: keep agent-run logs and integration-test logs for debugging and trend analysis; the repo contains example scripts to collect logs.

Example model config snippet (adapt the numbers to your provider in practice):

model:
  name: "your-model-name"
  # provider-specific fields here

Example retry snippet for tooling:

{ "retry": { "attempts": 3, "backoff_ms": 500 } }

Reference repo and artifacts: https://github.com/marketingskills/open-source-growth

What to do next (production checklist)

Assumptions / Hypotheses

• Initial setup time: 2–4 hours for a first working flow; ~3 hours is a reasonable single-run target.
• Team size for initial rollout: 1–3 maintainers will review the first agent PRs.
• Tuning examples to start with: model.max_tokens = 2048, retry.attempts = 3, retry.backoff_ms = 500.
• Operational gates: one automated PR per day (throttle), canary window = 24–72 hours, test pass threshold for blocking merges = 95%.
• Alerting and retention: alert on >10 failed runs/day; keep agent logs and integration-test logs for 30 days.
• Cost: modest API usage for a few runs; budget limits should be set in your configuration (amounts vary by provider and are not specified in the example repo).
• These numeric thresholds are starting recommendations you should validate against telemetry after 1–2 weeks. The overall skill categories and templates are drawn from the example repository: https://github.com/marketingskills/open-source-growth.

Risks / Mitigations

• Risk: low-quality or hallucinatory code in PRs

  • Mitigation: require CI tests and a human approval label before merge; block merges unless tests pass and a maintainer approves.

• Risk: secrets leakage

  • Mitigation: restrict PAT scopes, store secrets only in GitHub Secrets, and rotate secrets on a schedule.

• Risk: noisy PR volume

  • Mitigation: throttle to the configured rate (e.g., 1 PR/day), batch related changes, and validate in a canary repo first.

Next steps

• Harden workflows: require human approvals, pin model versions in skills/skills.yaml, and add explicit diff-size limits on automated PRs.
• Monitor impact: track PR conversion rate, demo-run success, and weekly changes; review results every 7 days and adjust thresholds in the assumptions list.
• Security & compliance: run a secrets-scan, add CONTRIBUTING and a CLA if you accept agent-generated content, and follow the patterns in https://github.com/marketingskills/open-source-growth.

Final reference: copy examples and checklist from https://github.com/marketingskills/open-source-growth and adapt skills/skills.yaml and .github/workflows/agent-apply.yml to your repository.

Implement marketingskills/open-source-growth agent skills to automate repo audits, README upgrades, demos and ecosystem PRs

Outline to use marketingskills/open-source-growth agent skills to automate repo audits, README upgrades, demo builds, launch packs and ecosystem PRs, plus quic…

https://aisignals.dev/posts/2026-06-27-implement-marketingskillsopen-source-growth-agent-skills-to-automate-repo-audits-readme-upgrades-demos-and-ecosystem-prs

(Weekly: AI news, agent patterns, tutorials)