Module reference

Fifty-five modules, seven families

A searchable source map of all 55 top-level modules. Open any card to see its runtime placement, responsibility boundary, public surface, and source directory.

55 source modules7 responsibility familiesOpen any card for its contract

Core loop

One step, and the things that make a step possible. Nothing here knows about a specific tool or model.

agent6,748

Defines executable agents and their lifecycle.

runtime735

Owns live Agent references, mailboxes, lifecycle state, and the event pump that advances them — and delegation, since one agent running another is a ref plus a job rather than a separate kind of agent.

session1,010

Defines shared invocation context and session-scoped project state.

message323

Defines provider-neutral conversation messages, multimodal content parts, tool calls, and function-call payloads.

response36

Defines the normalized `Response` and `ResponseType` returned across Agents, Tools, Skills, Connectors, and Environments.

protocol288

Defines typed agent-to-agent conversations over Runtime delivery.

Context & memory

What the model is shown, and what a finished run leaves behind.

model9,890

Provides the model registry, role-based selection, API-key pooling, and provider-neutral generation interface.

prompt738

Loads, versions, and renders HTML-native prompts and reusable prompt modules.

memory2,304

Stores and retrieves durable context derived from agent execution history.

trace1,158

Captures structured lifecycle events, persists them, and fans them out to subscribers.

trajectory576

Projects agent runs into reward-annotated, step-level training records and exports formats used by supervised fine-tuning or reinforcement learning pipelines.

Capabilities

The families of thing a model can be given. Each is a registry with the same shape.

tool10,523

Defines atomic callable capabilities backed by Python implementations.

skill15,645

Discovers and executes reusable, filesystem-backed instructions stored as `SKILL.md` plus optional scripts, references, resources, and examples.

environment5,277

Defines stateful execution environments and their callable actions.

connector6,906

Integrates external capability providers, primarily MCP servers.

plugins9,834

Outside services — OpenAI, Chroma, Tavily, YouTube, Composio — wrapped as plugins.

command798

Models user-facing commands, their contexts, and command dispatch.

knowledge336

Retrieval-augmented generation (RAG) over named knowledge bases.

code676

Runs a model-written program in its own interpreter and bridges the calls it makes back into this process, so a batch of tool work costs one turn instead of one turn per call.

Where work happens

Where file and process work actually happens — here, in a container, or on another machine.

sandbox3,025

Defines isolated command execution, managed sandbox processes, and staged project validation.

terminal657

Keeps a shell alive between tool calls, so a directory change, an activated environment, an ssh hop or a REPL survives the call that made it.

process760

Pure record/text transforms — the transform stage of a data pipeline.

kernel649

One Jupyter Server per project, and one kernel the agent, the Science REPL and JupyterLab all share.

e2b39

Provides the E2B backend integration point through `e2b_manager`.

ide334

Full VS Code (openvscode-server) in the browser, one container per gateway session, editing the same workspace the agent edits.

Orchestration

Work that is more than one step, and the limits on how far it may go.

workflow2,519

This package turns reviewable HTML into a dynamic multi-agent program.

canvas1,578

Visual flow editor.

task1,356

Models tasks and goals — their priorities, statuses and authority — loads task documents, and resolves CLI task input into normalized records.

job568

Runs work in the background and lets the agent collect it later — and holds reminders that come due — so a long command costs a step to start rather than a step spent waiting.

plan274

Holds a run to reading and reasoning until a person approves what it intends to do, so an approach is agreed before it is carried out rather than after.

constraint1,372

Defines enforceable runtime budgets and status reporting.

Evolution

How a component is created at runtime, versioned, judged, and removed again.

extension1,112

Manages generated extension manifests and their promotion into the active framework.

dynamic1,065

Loads generated Python source into controlled runtime modules and derives callable metadata such as parameters, argument models, and function-calling schemas.

version372

Tracks versions of evolvable framework entities and generates validated next versions.

capability112

Every callable Manager exposes `get_schema(name, action=None, format="json")`.

benchmark3,172

Provides versioned benchmark definitions and the manager used to load and execute them.

data767

Contains dataset adapters used by the Benchmark module, including AIME, GPQA, GSM8K, LeetCode, HLE, DeepWeb, and ProgramBench.

Infrastructure

The parts every other module leans on, and the one door the outside world uses.

gateway3,239

Defines the versioned boundary used by interactive clients to communicate with AgentEvolver.

config234

Loads Python configuration files into the global `config` object and validates assembled framework configuration before runtime initialization.

paths273

Single source of truth for the on-disk layout: one table declaring every path the framework writes, resolved through path_manager.

port183

Central port registry — names the framework's well-known default ports and hands out / records host ports, persisted to output/.runtime so bindings are de-conflicted and discoverable.

logger332

Provides the framework logger, log levels, display colors, and session-ID context used to correlate output across asynchronous operations.

permission719

Evaluates command and filesystem operations against explicit permission modes and policies.

hook2,387

Provides lifecycle interception points for tracing, compaction, registration, promotion, and other cross-cutting behavior.

utils1,708

Contains dependency-light helpers shared across modules: path assembly, file locking, concurrency, names, strings, URLs, screenshots, tokens, argument parsing, and plan models.

attachment212

Pins the bytes of an image the agent read and re-attaches it to every later request, so an image enters the model's context and stays there instead of vanishing with the step that read it.

deploy769

Deploys web artifacts from controlled source directories and records their URLs, health, resources, and lifecycle state.

conversation612

Lines of dialogue inside a project — their transcripts and their identity.

visual46

This directory contains dependency-free browser renderers for AgentEvolver's HTML-native artifacts.

The shape they share

Reading one module tells you how to read the rest. Four files, and the manager always spells removal the same way — unregister(name) — which is what lets one scope own registrations across several of them and take them all back out.

agentevolver/<module>/
  README.md      # YAML frontmatter + what this module owns  (gated)
  types.py       # the data and the contract — no I/O
  server.py      # the singleton manager: register / get / list / unregister
  __init__.py    # the public surface