- Home
- Skills
- Agents & Orchestration
- Kamari Burnett Skill Suite
Kamari Burnett Skill Suite
12 coordinated Skills for grounded task work, research, programming and frontend design. An evolving suite with two core Skills and task-relevant modules, updated from time to time.
Free
Works with the AI Agents you already use
Kamari Burnett Skill Suite
Example session with this skill installed
Use the suite to plan two weeks of customer research. I have eight hours per week, 20 trial users who did not reach activation, and a decision to make: change onboarding or improve the product’s core workflow. I have not interviewed anyone yet. Give me an executable plan and explain what evidence would change the decision.
- Read your context and instructions
- Compiled the kamari burnett suite
Two-week customer research plan
Use the 16 available hours to reconstruct why users stopped, observe their actual task, and test whether minimal guidance removes the obstacle. Aim for six 30-minute sessions, each followed by 15 minutes of notes. Six is a capacity target. The outcome is a justified next small test, not proof of activation lift.
Starting evidence: 20 trial users did not reach activation; no interviews have occurred. Neither onboarding nor the core workflow is yet established as the cause. This is a proposed plan; no research findings are claimed.
Define the decision
Write down the intended user, their task, and the useful outcome. Specify the existing activation event and measurement window; check whether they represent first value. If undefined, propose a task-specific first-value event and label it provisional.
Distinguish mechanisms: onboarding helps users understand, prepare, and begin; the core workflow performs the task and produces a usable result. Keep fit, timing, missing access, and measurement errors open as alternative explanations.
Budget and schedule
Each week totals eight hours. Move session slots to accommodate participants.
| Week | Work, in sequence | Minutes |
|---|---|---|
| 1 | Define decision, activation, and competing explanations | 45 |
| 1 | Review available trial records; map the 20 users | 60 |
| 1 | Invite, screen, and schedule participants | 45 |
| 1 | Prepare and rehearse the session guide | 45 |
| 1 | Three sessions, including immediate notes | 135 |
| 1 | Compare cases and identify the decisive uncertainty | 60 |
| 1 | Contingency: replacements, scheduling, missing context | 90 |
| 1 total | 480 | |
| 2 | Confirm participants or replacements | 30 |
| 2 | Prepare one minimal aid addressing week one's uncertainty | 60 |
| 2 | Three sessions, including immediate notes | 135 |
| 2 | Synthesize cases and examine contradictions | 90 |
| 2 | Write the priority decision | 60 |
| 2 | Specify follow-up validation | 45 |
| 2 | Contingency for no-shows or a decisive evidence gap | 60 |
| 2 total | 480 |
Use days 1–2 for preparation, days 3–5 for first sessions and synthesis, days 6–8 for the aid and remaining sessions, and days 9–10 for the decision. If records are unavailable, identify gaps for interviews; do not start a substantial analytics build.
Recruit
Invite all 20 neutrally; select six respondents covering relevant differences, rather than only the earliest replies. Where records permit, include users who never began a core task, began then stopped, and got furthest. Vary intended task and experience when relevant.
We are studying what people hoped to accomplish during their trial and what happened. Would you join a 30-minute conversation? We may ask you to show an example task if comfortable. We are evaluating the product, and participation is optional. Please share your availability and original goal.
Send one reminder after two business days; keep willing respondents as standbys. Do not promise an unapproved incentive. If fewer participate, report the missing perspectives and narrow the conclusion. Silence does not reveal why someone stopped. Respondents may differ from nonrespondents; this pool also cannot explain differences from activated users.
Run each session
Obtain consent for notes and separately for recording; recording is optional. Use sanitized task data.
| Minutes | Actions and prompts |
|---|---|
| 0–2 | Explain purpose, obtain consent, establish context |
| 2–9 | “What triggered the trial?” “What were you trying to accomplish?” “Walk me through your last attempt.” “Where did you stop, and what happened next?” |
| 9–12 | “How do you do this today?” “What would make the result useful?” Ask about deadlines, inputs, permissions, and alternatives |
| 12–22 | Observe the relevant task unaided first. Ask what they expect without supplying actions. After a meaningful block or about five minutes, offer minimal next-step guidance; mark the assisted phase and observe whether useful completion becomes possible |
| 22–27 | Inspect the outcome against their requirement. In week two, introduce the selected aid or concept only after the current-product attempt |
| 27–30 | Check your interpretation, invite corrections, and close |
Do not ask users to vote between your investment options. Connect feature requests to concrete tasks and consequences.
Week two's aid might be a checklist if setup appears decisive, or a paper sketch of a specific workflow change if capability appears missing. Cap preparation at one hour. A sketch tests understanding and relevance, not functional success. Assistance and repeat attempts introduce learning effects; improvement suggests a mechanism, not a causal effect estimate.
Record and decide
After each session capture: participant ID and context; source/prompt; direct observation or recalled account; unaided versus assisted outcome; suspected mechanism; rival explanation; and decision relevance. Preserve quotations faithfully or label paraphrases. Keep observation separate from interpretation.
At week one's review, choose the uncertainty most likely to change the decision. Seek contrary cases in week two. Compare by task and context, preserving contradictions. Counts describe observed cases, not population prevalence.
| Priority | Supporting evidence | Evidence that changes it |
|---|---|---|
| Onboarding first | Relevant users have a live task and access, become blocked by understanding/setup, then achieve useful results through the existing workflow with limited guidance | Understanding improves but useful completion still fails; assistance performs essential work; later independent use fails |
| Workflow first | Users understand the next step, but a material capability, effort, reliability, or output-quality problem persists despite guidance | Minimal guidance enables useful completion, or the requested change lacks a live task |
| Segment or sequence | Distinct groups have different barriers, or onboarding exposes a subsequent workflow failure | Further cases show one mechanism explains both patterns |
| Keep commitment provisional | Participation is narrow, evidence conflicts, observation is missing, or fit/access/measurement dominates | Obtain the specific missing observation that distinguishes explanations |
Judge recurrence across independent relevant cases, mechanism, severity, and effort—not a fixed interview threshold or majority vote. If both barriers occur, improving onboarding alone may move the stopping point without delivering value.
End with a short memo: chosen task/segment, smallest proposed change, case-linked support, strongest contrary evidence, uncertainty, and reversal condition. Specify the next test's activation denominator/window, independent task success, effort to first value, and quality/error guardrail. Check subsequent independent use with fresh relevant users where feasible. These sessions cannot establish uplift. Reverse priority if the targeted obstacle improves while the competing mechanism still blocks useful completion.
Connects securely to your tools. The creator never sees your data.
About this skill
A coordinated, evolving suite of 12 personal Skills
Use a connected AI agent to combine task planning, conceptual clarity, evidence integration, research, programming and frontend design. The two core skills guide information and effort allocation and evaluate whether the delivered result meets the user’s actual needs. Additional skills are selected as the task requires them.
Included Skills
- derive-on-demand — Determine the work and dependencies needed as a task develops.
- construct-entity-definitions — Clarify concepts, identity conditions, properties and supported distinctions.
- apply-practical-orientation — Pursue attainable objectives through action and feedback.
- apply-academic-orientation — Develop theoretical explanation and systematic understanding.
- dynamic-information-integration — Integrate evolving evidence, contributions and changing context.
- apply-local-information-effort-saving — Allocate information and effort to actions that can improve the outcome.
- exercise-self-directed-agency — Develop and revise goals, initiative and sustained pursuit.
- evaluate-output-usefulness — Evaluate and improve the actual result against the user’s needs.
- customer-research — Design or synthesize interviews, surveys and evidence about users and customers.
- frontend-design — Design or improve frontend interfaces and evaluate the rendered experience.
- program-structure-and-behavior — Understand, build, debug and verify programs and their behavior.
- construct-scientific-inquiry — Construct questions, hypotheses, research designs and valid inference.
Free access and ongoing updates
This suite is free to use through Agensi. It is an evolving collection, not a fixed one-time snapshot, and will be updated from time to time as its methods and coverage develop. No payment or monthly subscription is required. No fixed release schedule, quantity of updates, or guaranteed performance improvement is promised.
How it is used
Connect a compatible agent to Agensi and ask it to apply the suite to a concrete task. The suite provides instructions; it does not supply a model, external service accounts, software runtimes, paid API credits, or permission to act on your behalf. Tasks involving code, files, browsing or design still require the appropriate capabilities in your agent.
Agensi delivers the suite through its agent connection. This listing does not provide downloadable skill files or automatically change a user’s persistent system prompt or custom instructions. The creator’s companion prefix is being prepared separately and is not included as an exportable part of this listing.
Original module names and bundled third-party licence notices are retained. Results depend on the model, tools, task and evidence available. The worked example illustrates a research plan; it is not a completed customer study or a measured performance claim.
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.
Requires compatible host support for loading packaged references. The suite guides work but does not provide a model, runtime, external accounts or API credits. No guaranteed performance uplift or fixed update cadence. Does not modify persistent prompts. Skill contents are delivered through Agensi rather than downloadable files.
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
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Extra host tools are needed only when a task calls for them.
Permissions
The package contains instruction text, original UI metadata, SVG icons and licence notices; no executable scripts, credential collection or fixed external service integration. Read access to packaged references is required. Optional task execution can require user-authorized local files, shell, browser or network tools.