Back to Devoured - July 06, 2026
DEVOURED
Let Fable use its own judgement rather than dictating how it should work

Let Fable use its own judgement rather than dictating how it should work

AI llmagentscoding-agents Simon Willison
Simon Willison suggests delegating coding tasks to subagents running smaller, cheaper models rather than forcing top-tier models to perform every task.
What: Simon Willison advises using Claude Code to delegate tasks to lower-power subagents, specifically instructing them to use their own judgment for when to run tests or handle minor coding edits. He implemented this by adding a memory file at ~/.claude/projects/name-of-project/memory/delegate-coding-to-subagents.md to automate this delegation workflow.
Why it matters: This indicates a shift toward multi-agent orchestration, where users prioritize cost-efficiency by reserving high-capability models for complex reasoning while delegating mechanical tasks to smaller models.
Takeaway: Add a project-specific memory instruction to your Claude Code configuration to automatically offload mechanical coding tasks to a 'haiku'-class subagent.
Deep dive
  • Developers should avoid micromanaging agent behavior through overly specific constraints.
  • Prompting models to 'use their judgment' allows them to dynamically allocate resources based on task complexity.
  • Delegation patterns reduce token consumption by using smaller models for boilerplate and large models for core logic.
  • Claude Code's memory feature provides a persistent way to encode these project-specific architectural decisions.
  • Hierarchical agent workflows improve efficiency by separating implementation from architectural review.
Decoder
  • Subagent: An autonomous agent instance launched by a primary agent to execute a specific, isolated sub-task, often using a different underlying model configuration.
  • Fable: An internal Anthropic project or product designation referring to an advanced coding agent capability.
  • Claude Code: A command-line interface tool from Anthropic that allows agents to interact with files, git repositories, and shell commands directly.
Original article

One of the most interesting tips I got from the Fireside Chat I hosted with Cat Wu and Thariq Shihipar from the Claude Code team at AIE on Wednesday was to let Fable (and to a certain extent Opus) use their own judgement rather than dictating how they should work.

The example they gave was testing. You can tell Fable "only use automated testing for larger features, don't update and run tests for small copy or design changes" - but it's better to just tell Fable to use its own judgement when deciding to write tests instead.

Jesse Vincent just gave me a related tip to help avoid burning too many of those valuable Fable tokens in the few days we have left before the prices go up. Tell Fable to use other models for smaller tasks, applying its own judgement about which model to use.

I prompted Claude Code just now with:

For all coding tasks use your judgement to decide an appropriate lower power model and run that in a subagent

Claude saved this memory file in ~/.claude/projects/name-of-project/memory/delegate-coding-to-subagents.md:

---
name: delegate-coding-to-subagents
description: Simon wants coding tasks delegated to subagents running an appropriately lower-power model
metadata: 
  node_type: memory
  type: feedback
  originSessionId: 30068d78-43a9-4fb1-bb29-9799e18c526a
---

Stated by Simon on 2026-07-03: "For all coding tasks use your judgement to decide an appropriate lower power model and run that in a subagent."

Why: cost/efficiency — implementation work rarely needs the top-tier model; judgment, review, and synthesis stay with the main loop.

How to apply: when a task in this project is primarily writing/editing code, spawn an Agent with a model override (sonnet for substantive implementation, haiku for trivial/mechanical edits) and a self-contained prompt; review the result in the main loop before committing. Design, auditing, data synthesis, and anything judgment-heavy stays in the main model. See also [[project-goals]].

So far it seems to be working well. I'm getting a ton of work done and my Fable allowance is shrinking less quickly than before.

Explore all 67 articles in today's digest

Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi