The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted external data, such as earnings call transcripts and news articles from Kensho Grounding and S&P Global. It lacks explicit boundary markers or 'ignore' instructions to prevent potential prompt injection attacks embedded in this external content.
Ingestion points: Transcripts fetched via get_transcript_from_key_dev_id and external news findings from Kensho search queries.
Boundary markers: The instructions do not mandate the use of delimiters to isolate external data from the agent's core logic.
Capability inventory: The skill possesses the ability to execute shell commands (mkdir, cat, open) and perform local file writes, which could be targeted by an injection.
Sanitization: There are no specified procedures for escaping or sanitizing qualitative data retrieved from external sources before it is included in the report.
[COMMAND_EXECUTION]: The skill relies on shell commands for lifecycle management and local state handling. It specifically uses mkdir -p to initialize a temporary directory, cat to retrieve intermediate data, and open to launch the generated HTML report. This presents a potential command injection surface if the ticker symbol provided in $ARGUMENTS is not strictly validated before being used in file paths and commands.

earnings-preview-single