Struct WorkspaceManager 

pub struct WorkspaceManager { /* private fields */ }

Manages workspace lifecycle (create, list, load, status, delete).

Implementations

impl WorkspaceManager

pub fn ensure_agent_workspace(&self, agent_id: &str) -> Result<PathBuf>

Create or ensure a per-agent workspace exists.

Idempotent: if the workspace already exists, returns its path without modifying it. Creates the directory structure described in the module docs.

Directories are created with 0700 permissions on Unix for security.

pub fn create_agent_workspace( &self, agent_id: &str, template: Option<&Path>, ) -> Result<PathBuf>

Create a per-agent workspace from a template.

If template is None, falls back to ~/.clawft/agents/default/ if it exists, otherwise creates a bare workspace via ensure_agent_workspace.

pub fn delete_agent_workspace(&self, agent_id: &str) -> Result<()>

Delete a per-agent workspace by agent ID.

Removes the directory and all its contents from disk.

pub fn list_agent_workspaces(&self) -> Result<Vec<String>>

List all per-agent workspaces.

Returns a sorted list of agent IDs found under the agents directory.

pub fn link_shared_namespace( &self, exporter_id: &str, importer_id: &str, namespace: &str, ) -> Result<PathBuf>

Set up a cross-agent symlink for shared namespace access.

Creates a symlink from importer_id’s memory directory to exporter_id’s namespace directory. The symlink provides read-only access by default.

Security

• Validates both agent IDs to prevent path traversal.
• The exporter workspace must exist.
• The namespace directory must exist under the exporter.

impl WorkspaceManager

pub fn new() -> Result<Self>

Create a new manager, loading the registry from the default path.

The default registry path is ~/.clawft/workspaces.json.

pub fn with_registry_path(registry_path: PathBuf) -> Result<Self>

Create a new manager with an explicit registry path.

Useful for testing.

pub fn create(&mut self, name: &str, parent_dir: &Path) -> Result<PathBuf>

Create a new workspace.

Creates:

1. .clawft/ and subdirectories (sessions, memory, skills, agents, hooks)
2. .clawft/config.json with {}
3. CLAWFT.md with a starter template
4. Registers the workspace in the global registry

Returns the absolute path to the workspace root.

pub fn list(&self) -> Vec<&WorkspaceEntry>

List all registered workspaces.

pub fn load(&mut self, name_or_path: &str) -> Result<PathBuf>

Load a workspace by name or path string.

Returns the workspace root path if found.

pub fn status(&self, path: &Path) -> Result<WorkspaceStatus>

Get the status of a workspace at the given path.

pub fn delete(&mut self, name: &str) -> Result<()>

Delete a workspace by name.

Removes the entry from the registry but does NOT delete files from disk (that is the caller’s responsibility).