The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (HIGH): The setup instructions in references/mcp-setup.md recommend executing a remote script directly via shell: /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)". This 'curl-pipe-bash' pattern is a significant security risk if the remote source is compromised.
[EXTERNAL_DOWNLOADS] (HIGH): The skill uses npx -y octagon-mcp and npx skills add OctagonAI/skills to download and execute code at runtime. These sources (OctagonAI) are not within the explicitly trusted organization list defined in the security scope.
[INDIRECT_PROMPT_INJECTION] (MEDIUM): In SKILL.md, the workflow ingests external data (earnings call transcripts) to perform sentiment analysis.
Ingestion points: SKILL.md (Step 1: 'Analyze the overall sentiment... during 's latest earnings conference call.').
Boundary markers: Absent; the ticker content is interpolated directly into the analysis prompt.
Capability inventory: The skill uses an external MCP server (octagon-mcp) which performs the actual data retrieval and analysis.
Sanitization: Absent; there is no logic provided to sanitize or filter potential instructions embedded within the transcripts.
[CREDENTIALS_UNSAFE] (LOW): Both README.md and references/mcp-setup.md require the user to hardcode an OCTAGON_API_KEY into environment variables or configuration files. While standard for many tools, it increases the risk of accidental exposure in logs or shared environments.

earnings-conf-call-sentiment