The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute Python scripts. Specifically, it employs a Bash heredoc (<<'EOF') to securely pass JSON-formatted data from the agent's context to the create_proposal.py script. The use of a quoted heredoc prevents unintended shell expansion of the data being passed.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes untrusted external data in the form of sales call transcripts and client information. This data is used to generate expanded strategic paragraphs for the proposal.
Ingestion points: The skill accepts sales call transcripts as input in SKILL.md to extract project details and client information.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands within the provided transcript data.
Capability inventory: The skill has access to the Bash tool for shell execution, network access via the PandaDoc API in create_proposal.py, and file system write access in read_sheet.py (writing to the .tmp/ directory).
Sanitization: No explicit sanitization or filtering logic is present to validate the content extracted from transcripts before it is used in the proposal generation process.
[SAFE]: The skill communicates with well-known and trusted services, including the PandaDoc API (api.pandadoc.com) and Google Sheets API. These network operations are aligned with the primary purpose of the skill.
[SAFE]: Sensitive credentials such as the PandaDoc API key and Google OAuth tokens are managed using standard practices (environment variables and local token.json files). The instructions correctly guide the user to provide their own keys in a .env file.

create-proposal