CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

ainativedev/aidevcon-2026-nyc

AI Native DevCon 2026 New York — practical agent skills grounded in conference talks

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

application-template.mdskills/talk-bevin-agent-assisted-instrument-design/references/

Creative interaction experiment

This is a recommended application of Bevin's workflow, not a design specification from the talk.

Experience hypothesis

Person and context: Who is using the product, on which device and for which creative action?

Desired feeling: Describe the intended experience in the user's words.

Observable behavior: Name the action that should become easier or more expressive. For an instrument, this might be playing a chosen phrase without looking away or controlling a pitch bend reliably.

Current friction: Separate the reported symptom from guesses about its cause.

Prototype choice

Describe two or three approaches, the selected one and why it fits this product. State what existing behavior must survive.

Identify reusable components, such as an existing sound engine, separately from the new input surface. Do not invent platform APIs if they have not been inspected.

Real-context feedback

Record the build/version, device, interaction and tester. Capture what the person actually did, where control felt uncertain and which observations are subjective.

Include a short comparison with the previous version. Avoid declaring a feature universally intuitive because one tester liked it.

Verification path

Trace the full chain: physical input or event, interpretation, state change, output queue or service, and visible or audible result.

For each stage, state evidence available and the connection still unverified. Distinguish a passing helper test from an end-to-end effect. Use a clean build where stale artifacts could explain disagreement.

Decision

Choose keep, revise or discard, with a reason tied to the experience rather than the effort already spent writing code.

Record regression checks, unresolved uncertainty and the next bounded experiment. All proposed tests remain marked planned until observed. Preserve the user's actual release and commit workflow rather than adopting the speaker's personal settings automatically.

skills

talk-bevin-agent-assisted-instrument-design

tile.json