The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill uses a standard and secure environment variable (SENPI_AUTH_TOKEN) for authentication. Network operations in the client module (scripts/mcp_client.py) are directed to the vendor's official domain (senpi.ai), representing legitimate communication with the tool's backend infrastructure.
[COMMAND_EXECUTION]: The agent is instructed to run local Python scripts (scripts/sweep.py and scripts/score.py) that perform deterministic market analysis. These scripts do not process untrusted code or user-supplied strings as executable commands.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external market data from an MCP server. It mitigates potential injection risks by parsing data into specific numeric fields and providing the agent with a fixed glossary and strict narration rules that enforce "observation, not advice."
[DEPENDENCIES]: The skill bundles its own utility scripts (mcp_client.py, smartmoney.py) to minimize external dependencies and prevent supply-chain vulnerabilities associated with downloading third-party packages at runtime.
[OBFUSCATION]: The instructions and code were inspected for multi-layer encoding, homoglyphs, and hidden characters; no signs of obfuscation were detected.

senpi-signals