Documentation
¶
Index ¶
- Variables
- type LoadedSkill
- type Registry
- func (r *Registry) Delete(ctx context.Context, tenantID, skillName, version string) error
- func (r *Registry) Download(ctx context.Context, tenantID, skillName, version string) (io.ReadCloser, error)
- func (r *Registry) List(ctx context.Context, tenantID string) ([]SkillMeta, error)
- func (r *Registry) ResolveLatest(ctx context.Context, tenantID, skillName string) (string, error)
- func (r *Registry) Upload(ctx context.Context, tenantID, skillName, version string, zipData io.Reader, ...) error
- type SkillMeta
Constants ¶
This section is empty.
Variables ¶
var ErrSkillNotFound = errors.New("registry: skill not found")
ErrSkillNotFound is returned when a skill cannot be located in storage.
Functions ¶
This section is empty.
Types ¶
type LoadedSkill ¶
type LoadedSkill struct {
// Skill holds the parsed SKILL.md metadata and instructions.
Skill *skill.Skill
// Dir is the absolute path to the temporary directory containing the
// extracted skill files. The caller is responsible for removing this
// directory after use.
Dir string
// Entrypoint is the relative path (from Dir) to the main script
// that should be executed (e.g. "main.py", "main.js", "main.sh").
Entrypoint string
// HasRequirements is true when a requirements.txt file is present
// in the extracted archive, indicating Python dependencies.
HasRequirements bool
// HasPackageJSON is true when a package.json file is present in
// the extracted archive, indicating Node.js dependencies.
HasPackageJSON bool
}
LoadedSkill contains a fully validated skill ready for execution. It includes the parsed SKILL.md, the path to the extracted files on disk, the entrypoint script, and metadata about dependency files.
func LoadSkill ¶
func LoadSkill(ctx context.Context, reg *Registry, tenantID, skillName, version string) (*LoadedSkill, error)
LoadSkill downloads a skill archive from the registry, extracts it to a temporary directory, validates its contents, and returns a LoadedSkill.
The caller MUST remove LoadedSkill.Dir when done (e.g. via os.RemoveAll).
Validation includes:
- SKILL.md must exist and parse successfully
- A recognized entrypoint script must exist
- Zip entries are checked for path traversal (zip slip) attacks
type Registry ¶
type Registry struct {
// contains filtered or unexported fields
}
Registry provides skill archive storage backed by MinIO / S3-compatible object storage. Archives are stored as zip files keyed by tenant, skill name, and version.
func New ¶
New creates a Registry connected to the given S3/MinIO endpoint. It ensures the target bucket exists, creating it if necessary.
func (*Registry) Delete ¶
Delete removes a skill archive from the registry. It returns ErrSkillNotFound if the archive does not exist.
func (*Registry) Download ¶
func (r *Registry) Download(ctx context.Context, tenantID, skillName, version string) (io.ReadCloser, error)
Download returns a reader for the skill zip archive. The caller is responsible for closing the returned ReadCloser.
func (*Registry) List ¶
List returns metadata for all skill versions belonging to a tenant. It works by iterating over object prefixes under the tenant's namespace and extracting skill name, version, and upload timestamp from each matching object.
func (*Registry) ResolveLatest ¶
ResolveLatest returns the version string of the most recently uploaded version of a skill. If no versions are found, it returns ErrSkillNotFound.
func (*Registry) Upload ¶
func (r *Registry) Upload(ctx context.Context, tenantID, skillName, version string, zipData io.Reader, size int64) error
Upload stores a skill zip archive in the registry after validating that the provided data is a valid zip file (by checking zip headers). The archive is stored at {tenantID}/{skillName}/{version}/skill.zip.
Source Files