Compatibility and current limits#

These docs target framework 0.1.0 in this checkout. Preserve the exact Git revision for experiments; version numbers alone may not distinguish unreleased changes.

Execution support#

Capability

Current support

Core package

Python 3.11+ and Git

Local deterministic Quick Start

No model account, GPU, external dataset, or host CLI

Context-only workspace generation

macOS and Linux

Sandboxed learner programs

Linux with bubblewrap and permitted namespaces

Built-in methods

Prompt-only (the Codex learner on the prompt-only template), Codex, Claude, GEPA, SkillOpt, MWM

Workspace deployment

Codex, Claude Code, and Pi

Native plugin conversion

Claude Code, Codex, Pi, Hermes, with export restrictions

Stdio tool server

hermit serve DIR; normally started by the host

Generic manifest-driven standalone service lifecycle

Proposed; not implemented

Recipes can require newer Python versions and additional checkouts. The core package’s minimum version does not guarantee compatibility with every external benchmark.

Export compatibility#

Workspace and plugin paths have different capabilities. A target plugin may refuse context or roles that the same host accepts through project files. Read the conversion matrix. Use the output report to inspect omissions; --allow-lossy deliberately changes what the consumer receives.

Workspace deployment uses local source paths for tools. Conversion embeds an adapter copy and uses the installed framework’s stdio server. Neither command guarantees that arbitrary tool dependencies exist on another machine.

Known limitations#

  • agent.name is the current acting-agent selector. Legacy pipeline.run settings remain supported when no agent is named; function-based agents receive params and base when their signatures declare those keywords.

  • Trainer acceptance checks structure; performance selection belongs to the Evolver or your experiment protocol.

  • The sandbox covers launched learner programs, not the whole Python process or the acting pipeline.

  • Restricted network mode uses a proxy that a program can bypass by ignoring proxy settings.

  • Budget observation can overshoot with in-flight work; missing usage remains unknown.

  • Crash recovery restores the last completed framework step, not arbitrary external service state.

  • The evaluator’s solved_once threshold is fixed at score 1.0.

  • Dynamic per-call adapter context requires a custom consuming pipeline; ordinary exported instructions are static.

  • The proposed manifest serve section, health checks, standalone startup, and relay lifecycle are not supported.

Earlier command interfaces#

Older evolve / run instructions do not describe this interface. The current experiment commands are check, train, and eval; consumption uses deploy and convert. There is no documented automatic migration of old configurations or learner-state formats. Preserve an old experiment and create a separate 0.1.0 configuration, validating the pipeline, judge, manifest, and splits before spending on a new run.