From software problem to a clear next step.
Choose a topic and what you need to do next. Add a search phrase to find a specific problem.
Browse 20 topics
Explore the playbooks
1500 guides · Page 16 of 60
-
Triage an Oversized Agent Task Scope
Break an oversized agent request into a primary decision, bounded evidence slices, dependencies, and explicit stopping points before work expands further.
-
Prioritize Slices From an Oversized Agent Request
Order an oversized agent task by decision leverage, dependency, evidence readiness, urgency, and the cost of carrying unresolved scope.
-
Investigate How an Agent Task Became Oversized
Trace the original request, scope changes, evidence needs, and review points to identify where an agent task lost a workable boundary.
-
Verify an Agent Task Has a Workable Scope
Check that an agent task has one primary decision, finite inputs, explicit exclusions, reviewable output, dependencies, and a tested stopping condition.
-
Learn From Repeated Oversized Agent Tasks
Capture scope-growth patterns, slicing examples, decision ownership, and recurrence review so agent work stays finite while related questions remain visible.
-
Triage an Inconclusive Agent Result
Define the unanswered decision, competing explanations, available evidence, and smallest missing observation before treating an inconclusive result as failure.
-
Prioritize Follow-Up for an Inconclusive Agent Result
Rank unresolved agent results by decision consequence, uncertainty, evidence yield, collection cost, and whether a safe interim decision is available.
-
Investigate Why an Agent Result Is Inconclusive
Trace the decision boundary, evidence selection, comparisons, and competing hypotheses to identify why an agent cannot support a conclusion.
-
Verify an Inconclusive Result Is Honest and Useful
Check that an inconclusive agent result names its alternatives, evidence limits, supported claims, next discriminator, owner, and clear provisional boundary.
-
Learn From Repeated Inconclusive Agent Results
Capture unresolved-boundary patterns, alternative matrices, discriminating fixtures, and recurrence review so uncertainty stays precise and actionable.
-
Triage an Agent Task Missing Verification Steps
Define the intended outcome, observable checks, evidence owner, failure handling, and stopping condition before accepting an agent task without a verification plan.
-
Prioritize Verification Work for an Agent Task
Rank missing verification checks by consequence, failure likelihood, evidence cost, reversibility, and the decisions that depend on acceptance.
-
Investigate Why an Agent Task Lacks Verification
Trace the task brief, implementation claims, evidence, review handoff, and scope changes to find where verification obligations were lost.
-
Verify an Agent Task Has a Complete Verification Plan
Check that an agent task names observable outcomes, boundary cases, evidence, reviewers, failure handling, and acceptance criteria before it is marked ready for review.
-
Learn From Missing Verification Plans
Capture proof-plan gaps, acceptance examples, ownership, and recurrence review so agent tasks explain how their results will be judged.
-
Triage a Weak Agent Handoff Summary
Identify the receiving decision, missing evidence, context, uncertainty, ownership, and next action before a weak agent handoff creates repeated work.
-
Prioritize Repairs to a Weak Agent Handoff
Rank handoff gaps by receiver decision, repeated-work risk, evidence loss, urgency, and the smallest context addition that restores reviewability.
-
Investigate Information Loss in an Agent Handoff
Trace detailed agent work into its final handoff to locate lost scope, evidence, uncertainty, ownership, and next action that force repeated investigation.
-
Verify an Agent Handoff Summary Is Reviewable
Check that a handoff states decision, scope, evidence, conclusion, uncertainty, owner, next action, and stop condition for its receiving reader.
-
Learn From Repeated Weak Agent Handoffs
Capture handoff omissions, receiver questions, concise examples, ownership, and recurrence review so completed agent work stays usable after transfer.
-
Triage Unsupported Confidence in an Agent Result
Compare confidence language with claims, evidence, alternatives, and known limits so an agent result can be scoped before its certainty misleads review.
-
Prioritize Unsupported Confidence for Review
Rank confidence gaps by decision consequence, evidence weakness, scope, recurrence, and how quickly clearer uncertainty can prevent a misleading action.
-
Investigate Why an Agent Sounds More Certain Than Its Evidence
Trace claims, evidence selection, comparisons, instructions, and confidence rules to identify where an agent result gained unsupported certainty.
-
Verify Confidence Matches Agent Evidence
Check that an agent's certainty reflects claim type, evidence coverage, alternatives, freshness, and stated limits across representative workflow cases.
-
Learn From Repeated Unsupported Confidence
Capture certainty failures, evidence matched cases, wording standards, ownership, and recurrence review so agent confidence stays proportional to support.