v0.9.0 - Isolation (Containers, HF Sandboxes)

What's Changed

Container and HF Sandbox Support

To configure:

# hf sandbox
fast-agent -x --smart --url https://huggingface.co/mcp?bouquet=files -m "i want to set up an execution environment (hf sandbox) with my most recent dataset attached" --model codexplan 

# container
fast-agent -x --smart -m "configure a docker execution environment (ubuntu) named docker-env with a read only mount of the current working directory. make it the default"  --model codexplan

• Added named execution environments for shell/filesystem work: local, Docker, Hugging Face Sandbox, and custom adapters. Config now supports environment definitions and default_environment: in fast-agent.yaml.

• Added CLI selection with fast-agent go --environment <name> / -E, so shell- enabled agents can run in a configured environment.

• Reworked “environment directory” terminology into fast-agent home and workspace: - --env / --noenv became --home / --no-home. New --workspace controls the project root, with home defaulting to <workspace>/.fast-agent.

• Added environment-backed filesystem behavior: file tools now follow the active execution environment when it provides a filesystem, avoiding host fallback for remote/container shells.

• Added environment-aware skills discovery, so skills are scanned from the active environment filesystem for non-local runtimes. Map workspace volumes accordingly.

• Added examples and docs for Docker shell, Hugging Face sandbox, custom environments, named harness use, and artifact transfer.

• Improved attachment/media handling, including provider-fetchable links, byte-backed attachments, argument parsing, and clearer validation.

• Bump requests from 2.32.3 to 2.33.0 by @dependabot[bot] in #844
• [codex] Add agent environment runtimes by @evalstate in #840

Full Changelog: v0.8.3...v0.9.0