Hmbown/rlmagents

RLM agent harness - built on Deep Agents

What's novel

RLM agent harness - built on Deep Agents

Code Analysis

13 files read · 5 rounds

An agent framework implementing the Recursive Language Model pattern where an LLM works in a loop — loading data into sandboxed Python REPL contexts, executing code to analyze it, spawning recursive sub-queries, tracking evidence/citations, and producing cited final answers — built on LangChain/Lang

Strengths

Genuinely deep implementation with a well-designed AST-based sandbox, smart evidence pruning, context pressure compaction, and a complete recursive sub-query mechanism; clean middleware architecture with 27 profile-driven tools and consistent type safety throughout.

Weaknesses

Monolithic tool (73KB) and helper (47KB) files hurt navigability; the sandbox is explicitly best-effort and not hardened; test coverage could not be verified and the project is at alpha (0.0.5) with heavy external service dependencies.

Score Breakdown

Innovation

4 (25%)

Craft

82 (35%)

Traction

36 (15%)

Scope

86 (25%)

Signal breakdown

Innovation

Not Fork+1

Code Novelty+1

Concept Novelty+1

Craft

Ci+5

Tests+8

Polish+1

Releases+4

Has License+5

Code Quality+22

Readme Quality+15

Recent Activity+7

Structure Quality+5

Commit Consistency+5

Has Dependency Mgmt+5

Traction

Forks+6

Stars+20

Hn Points+0

Watchers+0

Early Traction+5

Devto Reactions+0

Community Contribs+5

Scope

Commits+8

Languages+5

Subsystems+10

Bloat Penalty+0

Completeness+7

Contributors+8

Authored Files+15

Readme Code Match+3

Architecture Depth+7

Implementation Depth+8

Evidence

Commits

688

Contributors

Files

435

Active weeks

TestsCI/CDREADMELicenseContributing

Repository

Language

Python

Stars

Forks

License

MIT