Explore a topic
Debugging and incidents
Find out what failed, keep the evidence and recover without making the problem worse.
Start with a practical guide
Notion or Vedokrok? Keep your runbook
Use the runbook your team trusts. Add a public method only when it fills a specific gap or gives you a check to run.
Read the story LongreadDebug an intermittent failure
Investigate a bug that appears only sometimes or a flaky test. Keep failed and successful runs, compare what changed and test competing explanations. Record what remains unknown.
Read the storyStart with a collection
Compare all collections on this topic
Find the fault without making the incident harder to understand
Turn vague technical failures into smaller tests, controlled incident work and system-level learning that another person can verify and reuse.
Find the failure without creating a second incident
Reduce a reproducible failure, locate its boundary and control the risks of retries, concurrent edits and recovery.
Catch weak signals before they become expensive errors
Use near misses, risky successes, repeated workarounds, leading indicators and layered safeguards to find weak points before harm repeats, while keeping reporting separate from verified improvement.
Design routine work so ordinary human slips do less damage
Separate slips from mistakes, reduce avoidable dependence on memory and vigilance, and use constraints, standardization, workload checks and recovery paths when the cost of an error matters.
Change live systems without turning one mistake into a large one
Make software, configuration and bulk-data changes easier to test, observe, retry, stop and recover before their blast radius becomes the main problem.
Operate AI agents as systems you can replay, verify and constrain
Add practical 2026 agent-engineering techniques for replay, evaluation, tool routing, computer use, scoped authorization and deterministic execution without turning conference claims into universal guarantees.
80 ideas
Replay from the failure checkpoint, not from memory
A reproducible failure is more useful than a fresh attempt that happens to succeed.
Read the idea ChecklistCapture semantic boundaries, not only raw logs
A log can be complete and still omit the boundary where meaning changed.
Read the idea ProtocolTest an intervention across a replay cohort
One rescued trace is a story. A cohort begins to look like evidence.
Read the idea TemplateTreat done as an evidence object
Done is not a green check mark. It is a set of claims someone must be able to inspect.
Read the idea PrincipleGive verification an independent evidence path
A second opinion is weak when it only rereads the first opinion.
Read the idea ProtocolTry to break the evaluator before trusting it
If the agent can game the judge, a higher score can mean a worse system.
Read the idea PrincipleSimulate users who behave like users
A simulator that helps the agent too much is grading a different job.
Read the idea PrinciplePut uncertainty around consequential eval differences
A two-point score gap is not a verdict until you know how noisy the estimate is.
Read the idea ProtocolMine traces for failure categories before changing prompts
A low score tells you that something is wrong. A failure taxonomy tells you where to work.
Read the idea ProtocolRoute to the cheapest model that still passes your task eval
Do not pay for intelligence the task does not use—but do not guess where the boundary is.
Read the idea ProtocolRetrieve tool schemas just in time
A tool can exist without occupying attention on every turn.
Read the idea PrincipleMeasure the active tool set, not the catalog size
The catalog is inventory. The working set is cognitive load.
Read the idea ProtocolRetrieve code before sending code
The cheapest token is the irrelevant file you never send.
Read the idea HeuristicTurn repeated browser work into persistent programs
Use the browser to learn the path; use code to stop relearning it every time.
Read the idea ProtocolPin the browser environment when comparing agent behavior
If the browser moves under your feet, you do not know what the agent change caused.
Read the idea ProtocolExchange broad credentials for task-scoped access
Do not hand an agent the master key when it needs one door for one job.
Read the idea PrincipleSeparate user-visible UI state from model-visible context
What the user can see and what the model must receive are different design decisions.
Read the idea ProtocolGive computer-use tasks a start state and success check
A screenshot is not a test case until you know where the task starts and what state counts as success.
Read the idea PrincipleLet the model propose; let deterministic code commit
Use the model for judgment; use the control plane for authority.
Read the idea ProtocolTurn a production finding into a reusable eval artifact
A failure is expensive twice if the system can forget it.
Read the idea ProtocolBuild the smallest failure you can reproduce
A bug you can summon is easier to interrogate than a bug you can only describe.
Read the idea ProtocolProbe the boundary between two components
Do not interrogate the whole chain when one boundary can answer where the bad value first appears.
Read the idea HeuristicBisect a large failure domain
When the haystack has structure, stop inspecting every piece of hay.
Read the idea HeuristicStart with what changed, not with certainty about it
The last change deserves an interview, not an automatic conviction.
Read the ideaExplore another topic
Project delivery
Estimate the work, agree on what counts as finished and check risky changes before you make them.
96 ideasCritical thinking and evidence
Test claims, separate evidence from interpretation and ask questions that can change what you believe.
70 ideasWorking with AI
Give AI a clear task and context. Check its work and set limits on what it can do.
133 ideas