[Feature]: Add first-class pip install support

[Drswith]

Problem or opportunity

Quantex can currently model managed installs for bun, npm, brew, cargo, and winget, but Python package installs are not first-class managed install methods. Python-based agent CLIs therefore have to be represented as unmanaged command hints, for example vibe currently exposes binaryInstall('pip install mistral-vibe') instead of a managed pip lifecycle method.

That makes pip-based agents less agent-friendly than npm/cargo/brew agents:

• installMethods cannot distinguish a pip package from a generic binary/manual command.
• capabilities and doctor do not report pip availability.
• install/update/uninstall flows cannot use a shared pip provider.
• structured output and schema cannot expose pip as a stable managed installer contract.

Desired outcome

Users and agents should be able to install, inspect, update, and uninstall supported Python-distributed agent CLIs through Quantex using a first-class pip managed install method when upstream documentation verifies pip as a supported distribution channel.

Expected direction:

• Add pip as a managed install type and package metadata target.
• Add a catalog helper such as pipInstall(...) for agent definitions.
• Add pip availability to qtx capabilities and qtx doctor output.
• Add a pip package-manager provider for install/update/uninstall behavior where the operation is reliable.
• Update structured schema/spec coverage so pip is part of the stable machine-readable contract.
• Migrate verified pip-based agent definitions, starting with Mistral Vibe if its pip install remains an upstream-supported path.

Scope and non-goals

In scope:

• Modeling pip as a managed installer for agent lifecycle operations.
• Cross-platform command planning for Windows, macOS, and Linux.
• Detection of an available pip entry point, with a deliberate fallback strategy if pip vs python -m pip needs to differ by platform.
• State recording, command rendering, structured output, and validation updates needed for the new install type.
• OpenSpec coverage for the behavior contract before implementation.

Non-goals:

• Publishing Quantex itself to PyPI.
• Managing arbitrary Python environments, virtualenv creation, pyenv/asdf policy, or uv tool installs as part of this issue.
• Expanding Quantex into a workflow orchestration platform.
• Changing the default Bun/npm self-upgrade behavior for Quantex.

Linked discussion or prior context

No prior GitHub discussion link. Local context observed during intake:

• src/agents/types.ts currently defines ManagedInstallType = 'bun' | 'npm' | 'brew' | 'cargo' | 'winget'.
• src/agents/definitions/vibe.ts currently represents pip install mistral-vibe as an unmanaged binaryInstall(...) command.
• openspec/specs/agent-catalog/spec.md and related update/capability specs will likely need a new OpenSpec delta.

Which durable artifacts should this likely create or update?

• OpenSpec

Done when

• An OpenSpec change captures the pip managed-install contract before implementation starts.
• pip is represented in agent package metadata, managed install types, install method helpers, installer capabilities, command rendering, and structured schema where applicable.
• qtx capabilities and qtx doctor report pip availability consistently with other managed installers.
• Install, update, uninstall, inspect/resolve guidance, and recorded install state handle pip-installed agents consistently with the existing lifecycle model.
• At least one verified pip-distributed agent definition is migrated from unmanaged command hints to the new managed pip method.
• Tests cover the new managed installer behavior, structured output/schema changes, and representative catalog usage.
• User-facing docs/specs are updated if the observable install surface changes.

Guardrails

• I understand this issue should promote durable outcomes into repo-native docs instead of adding new ad hoc root markdown files.
• I understand this issue should stay within Quantex's lifecycle CLI scope unless an explicit decision changes that scope.