MCP: Skills

Skills are reusable instructions you write once and make available to every AI tool connected through the MCP. A skill captures how your team does something, such as your screener conventions, how you pick incentive amounts, or your rules for grounding insights in transcript evidence, so agents follow your process instead of guessing at it. Admins manage skills under Settings → AI & MCP, the same page where you connect the MCP.

​

How skills reach your agents

Every skill you enable is advertised to connected AI tools automatically. When an agent connects, it sees a list of your enabled skills, each with its name and description, and loads the full instructions for whichever ones are relevant to the task at hand. You don’t have to tell people which skill to use; the agent picks based on the descriptions you write, which makes a clear, specific description the most important part of a skill.

​

Creating a skill

1. Go to Settings → AI & MCP and find the Skills section.
2. Click New skill.
3. On the Details tab, fill in:
   • Name: a short, human-readable label (e.g. Screener creation).
   • Slug: the identifier agents use to load the skill. It’s derived from the name automatically; you can override it. Slugs are normalized to lowercase and hyphens.
   • Description: one or two sentences on when an agent should reach for this skill. This is what the agent reads to decide whether the skill applies, so be specific about the situation it covers.
   • Instructions: the skill body, written in the rich editor (headings, bold, lists). This is the actual guidance the agent follows.
4. Click save. The skill is created disabled by default.
5. Toggle it on in the list when you’re ready for agents to see it.

Only enabled skills are advertised to agents. Disable a skill at any time to pull it back without deleting it.

​

Deleting a skill

When a skill is no longer needed, admins can delete it outright instead of leaving it disabled. Delete from either place:

• Skills list: click the trash icon on the skill’s row.
• Skill editor: open the skill and use the delete option in the slideout footer.

Confirm the deletion in the dialog. If the skill is enabled and gates any MCP tools, the confirmation also notes that those tools will become available without it. After deletion:

• The skill disappears from the Skills list and from list_skills over MCP.
• Any tool gated only by that skill is no longer gated.
• The slug becomes free to reuse for a new skill.

​

Requiring a skill before tool use

By default a skill is advisory: agents see it and use it when relevant. You can go further and require a skill, so that an agent must load it before it’s allowed to run certain MCP tools. This is how you enforce a process rather than just suggesting one. On a skill’s Settings tab:

1. Turn on Require before tool use.
2. Choose the scope:
   • Every MCP tool: the skill must be loaded before any tool runs.
   • Only the tools I pick: select the specific tools the skill gates (e.g. require your Screener creation skill only before screener-editing tools).

Required skills show a Required pill in the Skills list. When a required skill hasn’t been loaded, the gated tool call is softly declined and the agent is told exactly which skill to load first, so a well-behaved agent loads it and continues on its own, without the person driving the session having to intervene. Editing a required skill’s instructions re-arms the gate, so agents pick up your latest guidance rather than a stale version.

​

What makes a good skill

• Write a precise description. The agent chooses skills by description alone. “How we write screener questions to filter out fraud” beats “screener help.”
• Keep instructions actionable. Concrete rules, examples, and do/don’t lists outperform background prose.
• One skill, one job. Several focused skills are easier for an agent to select from than one long catch-all.

​

How to structure your skills

The most common question about skills is how big to make them. Should you write one comprehensive “Create a study” skill covering naming, incentives, screeners, and moderators? Or separate skills for each piece? Lean toward focused skills, split by decision rather than by step.

​

Why focused skills work better

An AI tool goes looking for a skill based on what someone asks it for. So your guidance only gets used if there’s a skill that sounds like the thing being asked about. That’s the trouble with one big “Create a study” skill. If your incentive rules are inside it, and a researcher asks “what should we pay people for this one?”, they won’t get your rules — they weren’t creating a study, so nothing pointed at that skill, even though the answer was sitting in it. Two other things follow from this:

• It’s all or nothing. There’s no way to read part of a skill. A long catch-all costs the AI the whole thing to answer a narrow question.
• You can only require a whole skill, not part of one. If your screener standards and your study setup live together, you either require both before any of those tools run, or neither. Keeping them apart lets you enforce the screener standard without slowing down study creation.

​

But don’t split too far

A skill can also be too small to ever get used. Take a naming convention for study titles: naming comes up while someone is setting up a study, never as a request on its own, so a skill just for naming would sit there unused. Rules like that belong inside the skill for the work they’re part of. Skills that never come up aren’t free, either. Every skill you enable takes up room in the list AI tools see, and that list has a limit, past it, skills stop being offered at all. A dozen clear skills work better than forty fragments.

​

A test that works

Ask: would someone ever ask for this on its own?

• “How should we name this study?” — No, that only comes up mid-setup. Put it inside study setup.
• “What should we pay participants?” — Yes, people ask that by itself. Give it its own skill.

If nobody would ask for it alone, it belongs inside a bigger skill. If they would, it deserves its own.

​

A shape that works for most teams

One skill per phase of research, each a few hundred words:

• Setting up a study — including your naming conventions and the order things happen in
• Writing screeners — your question standards and fraud checks
• Setting incentives — how you pick amounts
• Recruiting participants — sourcing, batch sizes, and approval before sending
• Participant communications — voice, required disclosures, and what never goes out unread
• Synthesis — how you ground claims in evidence

Each is short enough to be read quickly, specific enough that AI tools find it when it’s relevant, and self-contained enough to require on its own if you want to.

​

Pointing one skill at another

Focused skills don’t have to stand alone. If two of them overlap, write the connection into the instructions: If the researcher hasn’t set an incentive yet, load the adding-incentives skill before suggesting an amount. An AI tool can fetch another skill at any point, including partway through a task, so this generally works. Use it instead of copying the same rules into several skills, which drift apart the moment someone edits one copy. Two things to know:

• Use the slug, not the name. The slug is the short identifier next to a skill’s name, and it’s what AI tools use to look one up. Pointing at a title is vaguer and breaks quietly if someone renames it.
• A pointer is a suggestion, not a guarantee. It relies on the AI tool following your instructions. When something genuinely must be read first, require the skill instead — that’s the version that’s enforced.

​

Getting the most out of required skills

Requiring a skill is how you turn guidance into a standard. When a skill is required, an AI tool has to read it before it can use the tools you’ve picked. If it tries without, it’s turned away and told exactly which skill to read first, so in practice it goes and reads it, then carries on, without the researcher having to step in. Some things worth knowing before you start requiring things:

• Start everything as advisory. Turn a skill on, see how it actually gets used, and only require it once the guidance has proven right. Requiring a half-finished skill just makes everyone read a work in progress.
• Only require the things you can’t take back — sending invitations, messaging participants, deleting records. A draft you can edit afterwards doesn’t need it.
• Pick specific tools rather than “every tool.” Otherwise the skill gets read before even the smallest lookup.
• The AI only reads a required skill once per conversation, not before every action, so requiring one costs less than you’d think. Reconnecting starts a fresh conversation.
• Edits take effect right away. Change a required skill’s instructions and AI tools pick up the new version before they next use the gated tools, handy when you’re correcting a rule mid-project.
• Requirements add up. Every required skill is something the AI reads before it can use the tools you’ve gated. If several skills all apply to creating a study, it reads all of them before it can create one. Two or three is comfortable.

​

When one larger skill is the right call

Combine into a single skill when steps are genuinely inseparable, when doing one without the others produces something broken rather than merely incomplete, and when nobody would ever ask for one part alone. In that case the ordering is the guidance, and splitting it hides the very thing that matters. It also keeps requirements simple. If four required skills all apply to the same tool, the AI reads all four before any work starts, another reason to keep genuinely sequential work in one place.

​

Starter skills

​

Troubleshooting

Don’t see Skills under Settings → AI & MCP? Skills roll out alongside MCP access. Confirm with your Great Question account team that the feature is enabled for your account. A tool call keeps getting declined? Check whether a required skill is gating it. The agent is told which skill to load, so a capable AI tool should load it and retry automatically. If it doesn’t, ask it to load that skill by name (or slug) and try again. Agents aren’t using a skill you created? Confirm the skill is toggled on — new skills are disabled by default. If it’s enabled but still not picked up, tighten the description: agents choose skills by description alone, so a vague one gets skipped. Still need help? Contact us at support@greatquestion.co — median response time is 19 minutes during support hours.