The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the clawdhub CLI from the npm registry and fetches agent skill data (instructions and metadata) from https://clawdhub.com.
[COMMAND_EXECUTION]: Executes the clawdhub CLI tool to perform skill management tasks, including authentication, searching, and installation.
[REMOTE_CODE_EXECUTION]: The skill's primary function is to download and install external code/instructions (skills) from a remote registry. This creates a supply-chain vector where a compromised package or registry could lead to the execution of malicious logic within the agent's context.
[DATA_EXFILTRATION]: The CLI manages authentication tokens. The skill includes specific guardrails in SKILL.md and SAFETY.md intended to prevent these tokens from being exposed to the shell environment or piped to network-transmitting commands, acknowledging the underlying risk of exfiltration.
[PROMPT_INJECTION]: The skill identifies and attempts to mitigate Indirect Prompt Injection surfaces.
Ingestion points: External skills are ingested via clawdhub install and clawdhub update commands.
Boundary markers: The skill requires the agent to present full installation details (name, version, source) to the user and wait for explicit confirmation, regardless of adversarial framing or authority claims.
Capability inventory: Includes shell execution, file system modification, and network communication via the CLI tool.
Sanitization: Implements hash-based matching for skill updates to verify the integrity of the downloaded content.

clawdhub-hardened