Rated materials Radar Knowledge Safety Practical techniques What should I change next time? Practical notes and repeatable techniques for working with AI agents.
Search and filters How to isolate untrusted data in tool responses from model instructions, distinguish argument validation from return validation, and enforce controls at the execution layer.
Recipe External sources
95% Check what a tool response can authorise before the agent acts.
Report what the receipt measures before making a claim about savings.
Issue Skufosaur methodology
80% Report what the receipt measures before making a claim about savings.
Use compact tool replies that preserve errors, search scope and a way to continue.
Practice External sources
90% Use compact tool replies that preserve errors, search scope and a way to continue.
A repeatable baseline-versus-candidate protocol for quality, time, usage and cost.
Recipe Skufosaur methodology
90% A repeatable baseline-versus-candidate protocol for quality, time, usage and cost.
A short handoff note with the goal, acceptance criteria and the next verifiable step.
Recipe Skufosaur methodology
85% A short handoff note with the goal, acceptance criteria and the next verifiable step.
Acceptance checks for search, extraction and code changes.
Recipe Skufosaur methodology
90% Acceptance checks for search, extraction and code changes.
Decide when an agent should stop, retry or hand a decision back to the operator.
Recipe Skufosaur methodology
95% Decide when an agent should stop, retry or hand a decision back to the operator.
Nothing found yet Try another keyword or reset the filters. The collection is just getting started.
I label my own experiments separately. These materials do not represent measured Skufosaur effects. Confirmed results require explicit conditions, repeatable checks, and artifacts.