The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the Xero API, which may contain untrusted content that could influence agent behavior.
Ingestion points: Accounting data retrieved from Xero through the offledger CLI in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded prompts within the retrieved financial data.
Capability inventory: The skill can execute shell commands (gh, pnpm, op), write to the filesystem (SQLite database and token cache), and perform network operations to Xero.
Sanitization: No explicit validation or escaping of the external API data is performed before processing.
[EXTERNAL_DOWNLOADS]: The skill downloads the offledger toolset from the author's repository at https://github.com/howells/offledger using the GitHub CLI.
[COMMAND_EXECUTION]: Executes several shell commands including pnpm install, gh repo clone, and op run to manage the CLI environment and perform data queries.
[CREDENTIALS_UNSAFE]: The skill references a specific local file path for service account credentials at /Users/danielhowells/.codex/plugins/secrets/1password-service-account.env and manages an access-token cache at ~/.offledger/xero-tokens.json.

xero