Skip to content

Theme — use AI recursively

Apply the agent recursively to improve your collaboration workflows: drafting plans, refining instructions, reviewing code, and packaging tools.

A high-leverage pattern in computational research is applying the agent to refine the collaborative workflow itself:

Recursive task Research application Reference module
Pre-execution planning Formulate structured analytical proposals and identify open methodological questions Scoping work
Instruction maintenance Synthesize agreed rules into AGENTS.md and prune redundant guidelines Memory, Steering
Independent peer review Audit a prior session's analysis using a separate agent with a fresh context Verification
Tool codification Convert an interactive exploratory workflow into a parameterized script or custom skill Tools
Environment configuration Audit local dependencies and document environment setup in pyproject.toml Environment
Provenance tracking Draft informative Git commit messages and explain code diffs Version control

Always audit generated artifacts critically. Recursive workflows amplify productivity only when the human researcher maintains methodological leadership.

Scoping work · Memory · Verification · Tools · Steering