t3-p1-04authored3–5h

Context-window strategy experiment

requires: t3-p1-03

Spec

Give the agent a task requiring understanding of three or more interrelated files, and run it twice - once pasting all files in, once giving it tools to explore the repo itself. Record which produced fewer errors and your explanation why.

Why this matters

Context strategy is the highest-leverage variable you control and the least discussed. Running both arms on the same task is the only way to separate the strategy's effect from the task's difficulty.

Verify

No result yet. Run `learn verify t3-p1-04` locally, or push to trigger CI.

artifactartifact: experiments/context-strategy.md

unrun

judgejudge: experiments/context-strategy.md

unrun

Rubric

# t3-p1-04 — Context-window strategy experiment

**Task:** Give the agent a task requiring understanding of three or more interrelated files, and run it twice - once pasting all files in, once giving it tools to explore the repo itself. Record which produced fewer errors and your explanation why.

**Strong:** Directly answers every one of the review questions below in the affirmative, with specific evidence from the actual work, not a restatement of the task.

- Were both arms run on the same task?
- Is the conclusion supported by counts?

**Passing but weak:** The deliverable exists and technically satisfies the spec, but is generic, restates the task instructions instead of engaging with them, or only partially addresses the review questions below.

**Fails:** The deliverable is missing, or engages with none of the review questions below - e.g. it describes what was built without explaining the reasoning the task actually asks for.