Every feature, one page each

Features

Each feature below has its own self-contained page: a description, a working code example, and a Copy prompt button that copies a ready-to-paste AI implementation prompt for that specific feature.

Orchestration

APEX Synthesizer

APEX (Assembly & Precision Executor): confidence-scored, citation-aware multi-agent result synthesis.

Agent Spawning

Agents dynamically spawn sub-agents mid-loop, with bounded nested delegation.

Self-Improving Loop

Looping

Outer iteration loop for search-refine patterns, with self-consistency branching.

Verifier

Structured, LLM-graded critique drives targeted refinement, with an adaptive budget.

Procedural Memory

An ACE-inspired evolving playbook — agents accumulate what worked across runs.

Episodic Memory

A log of past runs, recalled by similarity and recency before each new one.

Input & Streaming

Multimodal Input

Attach images and PDFs to any user message with image()/pdf().

StreamPolicy

Choose exactly which event types a consumer sees.

FastAPI Integration

One call turns any Agent, Orchestrator, or WorkflowBuilder into an SSE endpoint.

Agent Behavior

Structured Output

response_model validates the final answer against a pydantic schema.

Human-in-the-Loop Hooks

AgentHooks.approve_tool can deny individual tool calls before they run.

Memory & Extensibility

Memory Providers

Pluggable context stores auto-injected into each LLM call.

Redis Memory Provider

A persistent, shared MemoryProvider backed by Redis.

Skills

Reusable capability bundles, self-evolving from converged loop runs.

Infrastructure

Retry & Fallback

Per-agent exponential backoff with model fallback chains.

Observability

OpenTelemetry spans for every LLM call, tool execution, and workflow step.

Logging

Toggleable structured/colour logs with dev/prod/verbose profiles and a shared get_logger().

CLI

deepcrew run workflow.yaml — declarative workflow execution.