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-arcolano-autonomy-and-trust/references/

Review capacity and trust experiment

Use this worksheet to improve one category of agent work. Its trial design is an implementation adaptation.

Diagnose the queue

Workflow and team: Observation period: Work generated: Work with a clear user or business purpose: Time to first review: Human review effort: Merged, abandoned, and unresolved work: Evidence for abandonment reasons:

Separate counts from estimates. Note whether changes are comparable in scope and risk. A growing queue may reflect prioritization, acceptance criteria, limited attention, or several causes.

Capture a repeatable judgment

Candidate work category: Examples engineers routinely accept: Counterexamples requiring deeper judgment: Acceptance criteria: Failure impact: Checks that support the decision: Uncertainty those checks do not resolve:

Use concrete properties instead of an undefined low-risk label. Name who maintains the criteria and how examples will change after failures.

Define the bounded trial

Current process: Proposed responsibility delegated: Safeguards retained: Exception behavior: Trial size or duration: Authorized decision maker: Return-to-review condition:

Do not change merge permissions through this worksheet. Specify evidence needed before considering any policy change within the user's authorized workflow. Distinguish a trial proposal from an executed trial and a tool recommendation from a decision to rely on its output.

Evaluate the result

MeasureBaselineTrial observationInterpretation and limits
Review delay
Human effort
Useful completion
Reversions or defects

Compare similar work. State whether apparent improvement could reflect easier tasks, changed staffing, or selection by engineers. Higher reversion among reviewed work may mean reviewers correctly concentrated on risk.

Decision: What has been learned: What remains unproven: Next smallest improvement:

Prefer a narrow supported conclusion over a promise of full autonomous delivery.

skills

talk-arcolano-autonomy-and-trust

tile.json