Alphabell.
Paper · Self-modifying agents

A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering

An agent autonomously discovers and writes new exact-computation operators for its own use, verifying them without ground truth to triple its answerable question set.

A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering
Figure from the paper, arXiv. Source
The loop

A frozen model runs nested loops to discover missing capabilities and author new domain-specific language operators for itself. It verifies these operators through consensus and conservation laws without external data, expanding its own executable library.

The loop
A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering
Paper · arXiv
  1. Frozen model attempts exact QA tasks
  2. Model authors new DSL operators
  3. Independent implementations verify operators via consensus
  4. Verified operators expand model's executable library
  1. Frozen model attempts exact QA tasks
  2. Model authors new DSL operators
  3. Independent implementations verify operators via consensus
  4. Verified operators expand model's executable library
↻ The improved system does the next round, and the loop turns again.

Why it is a road to recursion

Allowing an agent to safely expand its own executable vocabulary without human labels provides a scalable path to open-ended capability growth in exact domains.

Evidence

The inner loop roughly triples the answerable-question set, and the outer loop self-extends on disjoint starting sets.

dsloperator-inductionoracle-free

Related loops

More self-modifying agents →
Darwin Gödel Machine
PaperarXiv
Darwin Gödel Machine
  1. Coding agent edits its own code
  2. Scored child agents enter the archive
  3. Archived agents make later self modifications
  4. repeat
Self-modifying agentsUBC, Vector Institute, Sakana AI · 2025
Claude Code
ProjectAnthropic
Claude Code
  1. Claude Code writes its own code changes
  2. Changes become next Claude Code versions
  3. Release becomes harness for next development round
  4. repeat
Self-modifying agentsAnthropic · 2025
Recursive Self-Improvement through Collective Intelligence
ProjectarXiv
Recursive Self-Improvement through Collective Intelligence
  1. Model searches for multi-agent workflows
  2. Model selects the best executions
  3. Model trains on selected executions
  4. Model improves its base capabilities
  5. repeat
Self-modifying agentsSakana AI · 2026