The Agent Skills Directory

[SAFE]: The skill design prioritizes security through the use of transaction journaling (write-ahead journals), cryptographic verification (SHA-256 digests), and strict path traversal prevention for all filesystem operations.\n- [NO_CODE]: The skill package does not include executable scripts (Python or JavaScript), instead leveraging a high-performance Rust crate for capability discovery and provisioning.\n- [COMMAND_EXECUTION]: The documentation provides examples for using the get-fable CLI to perform non-mutating discovery, installation planning, and ecosystem health checks.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a potential attack surface through the processing of external capability manifests.\n
Ingestion points: The skill ingests catalog_layers from official, team, and local capability manifests as defined in the discovery procedure.\n
Boundary markers: All incoming manifests are validated against canonical JSON Schema definitions in schemas/eco/ to ensure structural integrity.\n
Capability inventory: The skill manages file copies, download extractions, and environment updates within a staging sandbox.\n
Sanitization: Execution contracts are compiled to filter tools and environment variables, exposing only the minimal set authorized for a specific task intent.\n- [EXTERNAL_DOWNLOADS]: The provisioning process involves downloading capability releases, which is secured by requiring immutable git commits or cryptographically hashed release archives, rejecting mutable references like main or HEAD.

fable-eco