Good AI Task

AI compatibility

AI can do the heavy lifting on Apollo query cleanup, but a developer needs to steer it.

Possible with caveats

Workable, but read the conditions.

Average across 1 submission.

58
avg / 100

The honest read

An AI code agent can meaningfully assist with auditing Apollo Client queries, generating fragment-based rewrites, and adding caching directives — but the audit requires deep understanding of the specific app's data access patterns, and the 30% payload reduction target demands iterative measurement against a live environment the agent may not fully control. The work is too context-dependent and multi-step to be fully automated without significant human oversight.

Aggregated across 1 submission.

The five dimensions

Repeatability

Low

Every codebase has unique query structures, component hierarchies, and data requirements. The agent must make bespoke decisions about which fields are truly redundant and how to restructure fragments — this is not a templated task.

Ambiguity Tolerance

Medium

The 30% payload reduction target is a concrete, measurable goal, which helps. However, what counts as 'redundant' queries or 'overfetching' requires interpretation of business logic and component behavior that isn't always self-evident from code alone.

Data & Tool Availability

Medium

The agent needs full read/write access to the codebase, the ability to run the app and measure network payloads, and ideally access to the GraphQL schema. Without a live environment or network profiling tools, the measurement step is blocked.

Error Cost

High

Incorrectly restructuring queries or applying aggressive caching directives can cause stale data bugs, broken UI states, or missing fields at runtime — issues that may not surface immediately and could affect production users.

Human Judgment Required

High

Deciding which fields are safe to drop requires understanding of product requirements and edge cases that aren't encoded in the code itself. Caching strategy also involves tradeoffs around data freshness that depend on business context a developer must own.

What an agent would need

  • Full read/write access to the React codebase and all Apollo Client query files
  • Access to the GraphQL schema (introspection endpoint or SDL file)
  • Ability to run the application and capture network payloads to measure before/after payload sizes
  • Understanding of which data fields are required by each component (ideally via component-to-query mapping or documentation)
  • A safe branch or staging environment to test changes without risking production

Best-matched agent type

Code Agent

The kind of agent this work would call for if it were a fit. For this task, it isn't.

Run your own fit check

Get a calibrated read on your specific task in under a minute.

Check a task