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-fowler-regenerative-software/references/

Bounded regeneration worksheet

This worksheet is Tessl's recommended application of Fowler's talk. Fill it with supplied evidence; use “unknown” for missing facts. It is an experiment design, not a claim that replacement has succeeded.

Component and surviving system

Name the smallest useful component and its callers. Describe why its blast radius is limited. List the interfaces, user experience, data ownership, and operational foundations that must remain stable. Name dependencies that are outside the visible code, including shared state and migrations. Explain why this component is replaceable rather than merely small.

Durable evidence

AssetRequired statementEvidence or gap
IntentWhat must remain true for users?Original decision or requirement
ArchitectureWhat boundary does the component uphold?Contract and caller evidence
EvaluationWhat observable result demonstrates the intent?Check independent of implementation
ProvenanceWhy was this behavior chosen?Person, decision, context

Distinguish intended behavior from an implementation accident. Do not promote every current quirk into a permanent requirement without a reason.

Side-by-side comparison

Record the original and candidate implementations, the shared evaluation inputs, expected outputs, state compatibility, and comparison method. Include one ordinary case, one failure case, and one behavior that the current code knows but the written requirements omit. These case categories are suggested here, not universal requirements from the talk.

For each divergence, classify the cause and record the decision needed. A passing implementation-specific test does not establish rewrite equivalence. Keep a separate place for checks that cannot yet survive a language change.

Feedback and decision

Define the production observation, the assertion it measures, and the component allowed to change when it fails. Use the user's actual target instead of copying the talk's 300-millisecond example. State evidence required for replacement, rollback conditions, and what will remain a thought experiment. Finish with the smallest next action that reduces an identified uncertainty.

skills

talk-fowler-regenerative-software

tile.json