The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes diagnostic shell commands to gather environment context for the debug bundle, including node -v, npm -v, and uname -a.- [DATA_EXFILTRATION]: The skill collects environment information, API status headers, and account quota metadata from api.juicebox.ai. It also accesses local application logs from logs/app.log. While these are non-whitelisted operations, they are consistent with the skill's primary purpose of creating a diagnostic support bundle. The skill includes explicit logic to mask the JUICEBOX_API_KEY before inclusion in text reports.- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local log files and external API responses to populate the diagnostic archive.
Ingestion points: Local file logs/app.log and JSON responses from api.juicebox.ai/v1/* endpoints.
Boundary markers: The skill does not use specific delimiters when appending log content to text files, but the output is intended for manual support review rather than immediate re-processing by the agent.
Capability inventory: Includes file system write capabilities (tar, redirection, fs.writeFileSync) and network access via curl.
Sanitization: Implements API key masking (${JUICEBOX_API_KEY:0:10}...) to prevent credential leakage in the generated logs.

juicebox-debug-bundle