The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the kweaver CLI tool via shell commands to perform administrative tasks on the platform, such as managing knowledge networks, agents, and data sources.
[EXTERNAL_DOWNLOADS]: The skill instructions include installing vendor-related Node.js packages like @kweaver-ai/kweaver-sdk and playwright via npm, as well as downloading and installing platform-specific extension packages using the kweaver skill install command.
[CREDENTIALS_UNSAFE]: The skill manages authentication tokens and database passwords. Commands like kweaver auth export, kweaver token, and kweaver ds connect involve printing or inputting sensitive credentials that could be exposed in command history or console output.
[DATA_EXFILTRATION]: The kweaver call (alias kweaver curl) command allows the agent to make network requests to arbitrary paths. This functionality could be used to transmit internal identifiers or authentication headers to external endpoints.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8) because it processes untrusted data from multiple sources.
Ingestion points: Data enters the context via kweaver agent chat (conversation history), kweaver bkn search (search results), kweaver ds import-csv (external files), and kweaver dataview query (database records).
Boundary markers: No explicit markers are used to delimit untrusted content within the instructions.
Capability inventory: The skill can execute shell commands and perform network requests via the restricted kweaver CLI.
Sanitization: There is no mention of sanitizing or validating data retrieved from the knowledge network or external sources before it is processed by the agent.

kweaver-core