Glossary · Term

agent skill

← all terms

Definition

Plain language

A documentation file an AI agent can consult at runtime to do a task better, like an employee handbook it may or may not open.

As stated in the literature

A markdown instruction document with trigger conditions and procedural guidance that an LLM agent loads at its own discretion; acts as a guidance prior over agent behavior rather than executable code, so failures can stem from bad instructions or from the agent ignoring good ones.

Also called: skill

Why it matters: It lets an agent improve at specialized tasks by reading guidance at runtime, but success depends on whether the agent actually consults and follows that guidance.

For example, an AI agent about to format a spreadsheet can open a skill file that explains the company's preferred layout before it starts editing.

Heard on the show

“It's attacker skill times defender fragility.”
Episode 202 — How Do You Know an AI Agent Actually Refused? Check the World, Not the Words

Mentioned in 62 episodes

  1. 202
    How Do You Know an AI Agent Actually Refused? Check the World, Not the Words
  2. 197
    Twin Problems Suggest AI Reasoning Gains Are Mostly Better Fact Recall
  3. 194
    How a Robot Builds a Debugging Notebook It Can Read, Edit, and Hand to Another Robot
  4. 193
    Freeze Most of the Network: Where RL Improvement Actually Lives in a Transformer
  5. 192
    A 32B Open Model Matched Frontier Systems By Learning to Take Notes
  6. 191
    How One Researcher Beat GPT-5.2 and Gemini 3 by Judging Their Answers, Not Improving Them
  7. 190
    The Skill Every AI Manager Is Missing: Handing Out Exactly the Right Keys
  8. 189
    Why Phone Agents Ace the Test and Crash on Your Actual Phone
  9. 188
    A Coding Agent Found a Hole in a Peer-Reviewed STOC Proof for Five Dollars
  10. 187
    An 8-Billion Agent That Beats Models 80 Times Its Size By Looking Things Up
  11. 186
    How a Frozen Model Went From 2% to 77% on Physics Puzzles — Without Retraining
  12. 184
    An AI Built an Undetectable Secret Channel, And Another AI Couldn't Find It
  13. 183
    Why You Can't Fine-Tune Foresight Into an AI Agent
  14. 180
    The Bug Where Smart Assistants Read a Fact and Still Forget It
  15. 175
    One Crosscoder Feature Flips a Stalling Chatbot Into a Working Agent
  16. 173
    The Free Step-Level Grader Hiding in Every RL Training Run
  17. 172
    One Bad Token Can Sink a Model's Math, And You Can Delete It
  18. 168
    When Turning Experience Into Code Makes Your AI Agent Dumber
  19. 167
    How Teaching an AI to Predict, Not Act, Made It a Better Actor
  20. 166
    A Router That Beats the Frontier Models It Calls
  21. 165
    A Free-Lunch Tweak That Lets a Tiny Agent Beat Frontier Giants
  22. 163
    Why Training Only on Perfect Solutions Cripples a Model's Reasoning
  23. 161
    A Robot That Plays Before You Give It a Job, And Why That Beats Retrying
  24. 160
    Training an AI to Take Its Own Notes, So Its Future Self Works Better
  25. 159
    Can a Coding Agent Run Its Own Robot Experiments Overnight, With No Human Resetting the Scene?
  26. 157
    When an AI Coding Agent Drives a Phone Through the Terminal, No Screen Needed
  27. 156
    Why More Human Demonstrations Made a Computer-Use Agent Worse
  28. 155
    Why a Flawless Demo Makes a Worse Computer-Using Agent, And the Fix
  29. 152
    Training a Model to Mean What It Says, And Why That Isn't the Same as Being Good
  30. 151
    Why More Experience Made This AI Agent Worse, And How to Fix It
  31. 147
    Agents Fail at the Body, Not the Brain: A Self-Rewriting Scaffold That Lifts a 9B Model 44 Points
  32. 142
    Training a Tiny Model to Run the Plumbing Between an Agent and the World
  33. 140
    When a Reasoning Model Says "Let Me Double-Check" After It's Already Decided
  34. 133
    How MiniMax Turned a Reward-Hacking Disaster Into Olympiad Gold
  35. 132
    The Agent Failed — But Did the Instructions Deserve to Be Followed?
  36. 126
    How Coding Agents Can Mine Their Own Failures Into a Self-Targeting Curriculum
  37. 120
    How an AI Agent Rewrites Its Own Tools, Without an Answer Key
  38. 112
    When an AI Agent Cheats Without Being Told: Inside the Meta-Agent Challenge
  39. 109
    An AI Got Caught Reading the Answer Key, And Why That Catch Matters
  40. 107
    How a Market of Crippled AI Agents Outscored One Unrestricted Model
  41. 106
    Giving Agents a Notebook Instead of New Weights: How ExpGraph Lets Frozen Models Learn
  42. 100
    How a Prompt Wrapper Lets a Frontier Model Play Poker Like an Expert
  43. 099
    How an Open-Book Trick Teaches a Model to Catch Its Own Mistakes
  44. 081
    When Reasoning Models Decide Before They Think: Detecting and Fixing Premature Confidence
  45. 080
    How a Two-Agent Trick Unlocked Large-Scale Training for Computer-Use Agents
  46. 079
    An Old Idea From Cognitive Psychology Reshapes How We Reward Reasoning Models
  47. 078
    Training a Markdown File: When LLM Self-Improvement Borrows the Discipline of Neural Net Training
  48. 072
    A Robot Made Graphene Without Help, And Caught Itself Hallucinating
  49. 064
    When Agent Memory Stops Being a Database and Starts Being a Skill
  50. 059
    Firefly's Inversion: Building Verified Tool-Call Training Data by Working Backward
  51. 054
    When Models Learn the Monitor Exists, the Reasoning Trace Stops Being a Window
  52. 053
    An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
  53. 052
    An Old Reinforcement Learning Tradeoff Sneaks Back Into LLM Agents
  54. 051
    Why Parallel Sampling Plateaus, And What Evidence Graphs Do Instead
  55. 049
    An AI Agent Reached for Root in Twelve Minutes, Without Being Attacked
  56. 047
    When Agent Benchmarks Lie: The Harness Problem in Open-Source AI
  57. 046
    When the AI Optimizer Edits the Grade Book: Why Harnessing Evolution Needs a Wall
  58. 043
    When 'This Is False' Doesn't Stick: Why Models Learn the Lie Anyway
  59. 027
    When AI Agents Build the Serving Stack: A Bet on Bespoke Infrastructure
  60. 021
    Ten Thousand Examples Beat the Full Industrial Pipeline for Search Agents
  61. 019
    When the Best Reward Model Trains the Worst Policy: Inside EvoLM
  62. 003
    How to Pick the Best of Sixteen Coding Agent Rollouts

Related concepts

Needle-in-a-Haystack Task Decomposition

Related terms

agent Markdown prior