- Home
- Skills
- Agents & Orchestration
- Using Agent Skills: Discovery and Invocation
More screenshots
Works with the AI Agents you already use
Using Agent Skills: Discovery and Invocation
The meta-skill for skill libraries: a standing contract for every task, a phase-based discovery protocol, a fit rubric, and rules for chaining skills into gated sequences.
$5
Using Agent Skills: Discovery and Invocation
Example session with this skill installed
I have a skill library and two things on my plate: users say checkout is slow, and next sprint I want to add promo codes. Where do I even start, and in what order?
- Read your context and instructions
- Compiled the using agent
ROUTING DECISION: two tasks, two different phases — routing them as one blended job is the first mistake this skill exists to prevent.
Task A, "checkout is slow": phase = debug. Nothing gets fixed before a diagnosis, so the sequence is diagnose → measure → test-first fix → review. Fit check on a performance-profiling skill: phase 2/2, trigger 2/2, do-not-use check clear, gate value 2/2 (its measurement gate catches the classic failure here — optimizing the layer that "feels" slow instead of the one that is). Assumptions surfaced up front: "slow" is undefined until measured (server time? page load? perceived?); the fix stays inside the checkout path unless evidence points elsewhere.
Task B, promo codes: phase = clarify/specify, not build. Nobody has said whether codes stack, expire, or apply before or after tax — building now would be building on guesses. Sequence: clarify → spec (gate: those three questions answered) → task breakdown → slice implementation → review → launch. Skipped deliberately: a database-migration deep-dive (existing schema already has a discounts table) and a full security review cycle (no new payment surface; a targeted review of the discount logic is scheduled instead).
Order across both: A first — it is user-facing pain with a short diagnose-to-fix loop; B's clarify step can run in parallel since it needs the user, not the codebase. Named sequence, shared: "A: diagnose → fix → review. B: clarify → spec → build (next sprint). Gates and skipped skills as noted."
Connects securely to your tools. The creator never sees your data.
About this skill
A skill library only pays off if the right skill gets invoked at the right moment, in the right order, and its verification step actually runs. This skill is the routing layer. It opens with the standing contract that applies to every task with or without a domain skill — verify don't assume, surface assumptions early, stop when confused, push back on bad ideas, enforce simplicity, hold scope — then gives a discovery protocol that matches the task's phase of work (clarify, specify, plan, build, debug, review, ship) to the catalog instead of matching keywords. A fit rubric scores a candidate skill before invocation, including a do-not-use check that treats an exclusion match as a negative result. For larger work it composes skills into named sequences with gates: each skill's verification is the entry ticket to the next, skipped gates are named rather than silent, and parallel tracks run where they are genuinely independent. It closes with conflict rules, edge cases (emergencies, stale skills, deliberate overrides), and a pre-delivery quality gate.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
- It routes work; it does not contain the domain workflows themselves. You still need the skills it points to (or the standing contract fallback).
- The fit rubric assumes skill descriptions honestly state their do-not-use conditions; vague descriptions score worse and should be rewritten, not routed around.
- For trivial tasks the correct output is "no ceremony needed" — if you want a process document for a one-line fix, this skill will decline to invent one.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Permissions
No special permissions declared or detected