Implicit Core

Implicit is the experience layer for AI agents.

Virtualize large agent environments. Materialize only the state each experience actually needs.

Implicit Core is AI agent experience/environment infrastructure from zeitcow/implicit-core, distributed as implicit-ai under Apache-2.0. Current stable: 1.0.1.

pip install implicit-ai

Website and docs · PyPI · 60-second quickstart · Measured evidence · Connect a coding agent · Install the Codex integration skill

Why virtualize an experience?

In agent training and evaluation, the possible environment can be much larger than any individual episode. A warehouse may contain millions of orders; processing one order needs only its inventory and policy records. Instead of eagerly constructing the complete environment for every episode, an Implicit experience addresses a coordinate in a versioned environment. Your adapter supplies the initial pages and loads further required state on demand, including declared dependencies.

An experience is one addressable interaction: instruction, required state, execution and evaluation. The address identifies the environment version and coordinate. Your adapter chooses pages; your existing agent executes; your native evaluator verifies the result. Core records execution state, provenance and explicit recovery.

Immutable source versions, fixed seeds and deterministic adapters, state sources and agents support reproducible episodes. External services and effects need application-owned control or reconciliation; Implicit does not make arbitrary programs deterministic. Architecture explains the conditions.

This works at the logical experience/environment-state layer. VM or container snapshots can supply an execution sandbox; Implicit determines which addressed state is needed inside an experience. FAQ distinguishes this from filesystem paging, dataset streaming, prompt compression and lazy MCP tool discovery.

Use Implicit for large separable state, repeatable identities, expensive environment construction or durable execution evidence. Measure overhead for small, already lazy or mostly accessed environments.

Try it in 60 seconds

Python 3.11+; zero third-party runtime dependencies. In a fresh virtual environment:

pip install implicit-ai
implicit --version
implicit demo
implicit benchmark

The public toy compares eager and selective serialized state, verifies the shipping result and prints a provenance hash. Its output is its own workload measurement.

The distribution is implicit-ai; the import is implicit. Use an isolated environment because the unrelated implicit recommendation library shares that import namespace. Installation includes Windows setup and verified release artifacts.

Measured evidence and its limits

In the preserved rc1 native systems population, 155/155 comparable cases preserved equivalent state, tool behavior and reward.

Measurement	Result
Aggregate retained serialized/materialized-state reduction	93.88%
Mean case reduction	94.40%
Median case reduction	99.55%

Serialized/materialized bytes are not RAM. The measurements belong to rc1, not a new 1.0.0 native replay or the toy/MCP demo. The public numeric summary supports arithmetic verification; restricted native replay assets are not shipped. Methodology, measured latency tradeoffs, hashes and limitations.

Keep your existing stack

Keep your agent, learner, framework and evaluator. Build an adapter in your project using ordinary Python protocols:

Universe proposes versioned addresses
  -> Environment plans required pages
  -> ResourceSource loads pages as needed
  -> existing agent executes
  -> native Evaluator verifies
  -> Core records provenance and recovery state

Adapter contract, agent integration prompts and three public adapter shapes show the mapping. With the public repository downloaded and the package installed:

python -I examples/core_adapters.py

No allocator or new learner is required. Default selection preserves your proposed order. Core does not establish improved learning, general speedups, allocator superiority or universal infrastructure guarantees.

Let a coding agent try Implicit

The package includes eight bounded local stdio MCP tools for synthetic addressing, paging, provenance and validation. MCP quickstart gives Codex, Claude Code, VS Code and Cursor configurations. Real environment adapters use the Python SDK.

Repository plugin packages integration guidance and local MCP configuration. Public directory acceptance and hosted ChatGPT connectivity are separate; see the dated ecosystem status.

Documentation

• Quickstart, installation and architecture
• Adapters, configuration and agent integration
• Benchmarks, citation guide and FAQ
• Local MCP, MCP quickstart and ecosystem status
• Security and privacy, troubleshooting and contributing
• Agent commands, release history and public adoption measurement

Demo, benchmark and local MCP make no outbound connections. Python adapters are trusted application code and may use your services. There is no product telemetry. Journals may retain application data; see SECURITY.md.

Implicit Core 1.0.1 is licensed under Apache-2.0. Licensing inventory and NOTICE describe included assets.