The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes npx to fetch the @useclawlink/cli package from the npm registry during the setup and execution phases.
[COMMAND_EXECUTION]: The agent is instructed to perform operations like login, connect, and run by executing shell commands through the ClawLink CLI.
[REMOTE_CODE_EXECUTION]: The use of npx without pinned versions facilitates the download and immediate execution of remote code from the npm registry.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Coda documents, which serves as a potential vector for indirect prompt injection if the content contains malicious instructions.
Ingestion points: Content is ingested through actions such as coda_get_a_column and coda_get_content_types as defined in SKILL.md.
Boundary markers: The skill does not provide explicit instructions or delimiters to help the agent distinguish between data and instructions when processing Coda content.
Capability inventory: The skill has the capability to execute shell commands and perform authenticated network requests (writes) to the Coda API.
Sanitization: There are no specified sanitization or validation protocols for data retrieved from Coda before it is interpreted by the agent.

coda