# A method for your next question

Six optional, editable workflows for connecting SocialToAI and researching public social evidence with clear budgets and sources.

Canonical: https://socialtoai.com/skills/

## Choose the workflow

socialtoai-research: route a general research question to suitable platforms and a bounded sequence of calls. socialtoai-setup: connect your client, test free discovery and configure requested preferences.

socialtoai-benchmarks: compare relevant creators with compatible metrics. socialtoai-viral-analysis: inspect a post and its audience response while separating evidence from causal hypotheses.

socialtoai-pain-points: collect problems, workarounds and counterexamples. socialtoai-topic-research: turn sourced audience questions into a topic shortlist.

Skills are optional instructions for compatible clients. The same core tools work without them; installing a Skill does not guarantee that a model will follow every instruction.

[Read research SKILL.md →](https://socialtoai.com/skill-files/socialtoai-research/SKILL.md)

[Read setup SKILL.md →](https://socialtoai.com/skill-files/socialtoai-setup/SKILL.md)

[Read benchmarks SKILL.md →](https://socialtoai.com/skill-files/socialtoai-benchmarks/SKILL.md)

[Read viral-analysis SKILL.md →](https://socialtoai.com/skill-files/socialtoai-viral-analysis/SKILL.md)

[Read pain-points SKILL.md →](https://socialtoai.com/skill-files/socialtoai-pain-points/SKILL.md)

[Read topic-research SKILL.md →](https://socialtoai.com/skill-files/socialtoai-topic-research/SKILL.md)

## Install a Skill

Run the command in your project. It installs the general research Skill from the public MIT-licensed repository github.com/yuwen-cool/socialtoai-skills. Replace its name with another workflow, or choose another supported agent. To install offline, download version 0.1.0, extract it, and pass ./socialtoai-skills instead of the repository name.

Review the installation prompt and the destination files. Then reopen your client and confirm it discovers the Skill. Connect MCP separately using the quickstart; Skills contain no credentials.

```
npx skills add yuwen-cool/socialtoai-skills --agent codex --skill socialtoai-research
```

[Skills repository ↗](https://github.com/yuwen-cool/socialtoai-skills)

[Download six Skills (.zip) →](https://socialtoai.com/downloads/socialtoai-skills-0.1.0.zip)

[Connect MCP →](https://socialtoai.com/docs/quickstart/)

[Skills installer documentation ↗](https://github.com/vercel-labs/skills)

## Make the method yours

Copy one workflow into a separately named local Skill. Change its folder and frontmatter name together, then add your industry's vocabulary, a competitor shortlist or editorial preferences. Keep your fork separate so an official upgrade cannot replace it.

Keep credentials out of Skills. The package has no telemetry or upload script. Terms you ask your AI to search will be sent in that search, so include only information you intend to use with the connected tools.

## Upgrade or remove

To upgrade, run the install command again; for a downloaded package, extract the new version first. Review replacement files and confirm the version in SKILL.md. Save custom edits in a separately named fork before replacing official files.

To fully uninstall the named Skill across this project, run the remove command without an agent filter. Clients can share .agents/skills; removing just one agent link may leave it discoverable. Verify with npx skills list --json and reopen the client. Removing a Skill does not disconnect MCP or revoke its Key. Remove the MCP connection in your client, or rotate a compromised Key in Access.

```
npx skills remove socialtoai-research
```

## Know what the method can establish

The bundle targets Contract v1 and includes platform references generated from the same cards and prices as this site. Check current tool descriptions before relying on a filter or price.

Budget, source citation, fallback disclosure and stop rules are part of every research workflow. They are instructions, not evidence of model compliance or guaranteed research quality. Public samples do not establish market share or private ranking algorithms.

[Prices →](https://socialtoai.com/pricing/)

[Errors and billing →](https://socialtoai.com/docs/errors/)