Checking an agent's work
Treating an agent's report as a claim to check, not as proof - asking for the real evidence (the command and its output), reading the diff, running it yourself or having a fresh agent review it - and better still, giving the agent tests or scripts so it can check its own work in the loop, against a definition of done agreed in advance.
Depth 1 · Foundation
You can explain why an agent's "done" can be wrong without anyone lying, and check its work through evidence rather than its summary.
Sign in to prove what you know with a short quiz, see how deep you've gone, and earn XP for it.
Sign in with GoogleDepths of understanding
- 1
Foundation
You can explain why an agent's "done" can be wrong without anyone lying, and check its work through evidence rather than its summary.
Your path here
What's still between you and Checking an agent's work, depth 1, in the order you can learn it.
- 1Language modelDepth 1 · Foundation · Language Models
- 2AI agentDepth 1 · Foundation · AI Agents
- 3TokenDepth 1 · Foundation · Language Models
- 4ProbabilityDepth 1 · Foundation · Probability
- 5Next-token predictionDepth 1 · Foundation · Language Models
- 6Stateless modelDepth 1 · Foundation · Language Models
- 7Context windowDepth 1 · Foundation · Language Models
- 8System promptDepth 1 · Foundation · Language Models
- 9Tool useDepth 1 · Foundation · AI Agents
- 10Agent loopDepth 1 · Foundation · AI Agents
- 11BeliefDepth 1 · Foundation · Epistemology
- 12KnowledgeDepth 1 · Foundation · Epistemology
- 13TestimonyDepth 1 · Foundation · Epistemology
- 14Checking an agent's workDepth 1 · Foundation · AI Agents
Where it fits
Builds on
This concept