agentmods.dev

optillm CLAUDE.md

Repository instructions for OptiLLM, a service that sits between applications and language models and tries different reasoning or sampling methods before returning an answer. It provides an OpenAI-compatible API, meaning programs can call it using the same general interface as OpenAI services.

In plain words
What is it for?
Use it when working on the proxy server, local model execution, inference strategies, answer checking, request routing, or the project's setup and development commands.
Why use it?
They explain the project's main parts and the available methods for improving answers to coding, logic, and maths questions. This gives the coding agent a map of where inference, routing, and optimization code belong.

Instructions file

4.3k 1mo ago A scan 1,227 tokens original Apache-2.0
View the source file algorithmicsuperintelligence/optillm
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/algorithmicsuperintelligence/optillm/claude-md
Clone the repo
git clone --depth 1 https://github.com/algorithmicsuperintelligence/optillm
Per session 1,227 This file is loaded in full into every session.
When invoked 1,227 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01227 $0.01227
Opus 5 $0.00613 $0.00613
Sonnet 5 $0.00245 $0.00245
Haiku 4.5 $0.00123 $0.00123

Measured yesterday against content hash 6efefe717d1c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

optillm CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 134 lines

How it starts

The opening of the file, as written. The whole thing — 134 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

OptiLLM is an OpenAI API compatible optimizing inference proxy that implements state-of-the-art techniques to improve accuracy and performance of LLMs. It focuses on reasoning improvements for coding, logical, and mathematical queries through inference-time compute optimization.

Core Architecture

Main Components

  1. Entry Points:

    • optillm.py - Main Flask server with inference routing
    • optillm/inference.py - Local inference engine with transformer models
    • Setup via pyproject.toml with console script optillm=optillm:main

  2. Optimization Techniques (optillm/):

    • Reasoning: cot_reflection.py, plansearch.py, leap.py, reread.py
    • Sampling: bon.py (Best of N), moa.py (Mixture of Agents), self_consistency.py
    • Search: mcts.py (Monte Carlo Tree Search), rstar.py (R* Algorithm)
    • Verification: pvg.py (Prover-Verifier Game), z3_solver.py
    • Advanced: cepo/ (Cerebras Planning & Optimization), rto.py (Round Trip)

  3. Decoding Techniques:

    • cot_decoding.py - Chain-of-thought without explicit prompting
    • entropy_decoding.py - Adaptive sampling based on token uncertainty
    • thinkdeeper.py - Reasoning effort scaling
    • autothink/ - Query complexity classification with steering vectors

  4. Plugin System (optillm/plugins/):

    • spl/ - System Prompt Learning (third paradigm learning)
    • deepthink/ - Gemini-like deep thinking with inference scaling
    • longcepo/ - Long-context processing with divide-and-conquer
    • mcp_plugin.py - Model Context Protocol client
    • memory_plugin.py - Short-term memory for unbounded context
    • privacy_plugin.py - PII anonymization/deanonymization
    • executecode_plugin.py - Code interpreter integration
    • json_plugin.py - Structured outputs with outlines library

Development Commands

Read the full file on GitHub · 134 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 134 lines · 1,227 tokens per session scan A 6efefe717d1c

Subscribe to this mod's changes

optillm CLAUDE.md is an instructions file published in the GitHub repository algorithmicsuperintelligence/optillm (4,257 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 1,227 tokens to every session, about $0.0061 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Backend and APIsData and AIPlanningagentagentic-aiagentic-workflowagentsapi-gatewaygenai
Related

Other instructions, from other repositories

LeAgent AGENTS.md

Instructions for vixues/LeAgent, covering leagent development guidelines, project overview, architecture, execution topology (one kernel, many ingresses) and surface map (where things live).

vixues/LeAgent · 6,799 tokens

grix AGENTS.md

Instructions for askie/grix, covering grix agent guide, repository boundaries, required workflows, shared agent configuration and cross-component contracts.

askie/grix · 520 tokens

rn-dev-agent AGENTS.md

Instructions for Lykhoyda/rn-dev-agent, covering repository guide for agents, repository map, editing rules, architecture rules and supported node runtimes.

Lykhoyda/rn-dev-agent · 5,241 tokens

grix CLAUDE.md

Instructions for askie/grix, a project described as: Grix : Work with agents like talking to people.

askie/grix · 5 tokens

clarion CLAUDE.md

Claude Code instructions for celanthe/clarion, covering clarion — agent voice manager, what it does, architecture, running it and ui only (edge tts, zero config).

celanthe/clarion · 1,385 tokens

upwork-agent CLAUDE.md

Claude Code instructions for citizenweb3/upwork-agent, covering upwork agent, setup (for a new machine), prerequisites, step 1: install dependencies and step 2: configure environment.

citizenweb3/upwork-agent · 2,731 tokens