Scribo Skill has 6 tools with clear verb-noun naming (list_, request_, redeem_, create_, get_, download_). All tools have descriptions (avg 120 chars, within 10-1024 range). However, input schemas are minimal or absent for most tools. list_jurisdictions has empty input {}; request_verification, redeem_verification, get_invoice, and download_invoice have single-parameter schemas with basic descriptions; create_invoice has a complex 'payload' object parameter but no documented output schema. No tool documents its return type or structure. Parameter descriptions exist but lack format constraints, ranges, or validation guidance. Error handling is mentioned in descriptions (e.g., 'verification_required status') but no recovery guidance is provided. The skill lacks output schema documentation entirely, LLMs cannot plan downstream calls without knowing what fields to expect. Naming is strong; schema and output documentation are weak.

• Document output schema for each tool. Example for create_invoice: 'Returns {invoice_id: string, document_id: string, format: string, download_url: string, download_url_expires_at: ISO8601, validator_summary: {valid: boolean, errors: array}}'.
• Add format constraints to parameters. E.g., request_verification email: 'Valid email address (RFC 5322)'. redeem_verification code: '6-digit numeric code from verification email'.
• Define create_invoice payload schema explicitly. Document required fields (sender, recipient, line_items, currency) and optional fields (idempotency_key). Provide a minimal example structure.
• Add error recovery guidance to tool descriptions. E.g., create_invoice: 'If verification_required, call request_verification() first, then redeem_verification() with the code, then retry create_invoice with the token.'
• Document idempotency behavior. create_invoice supports idempotency_key, state that repeated calls with the same key return the same invoice, enabling safe retries.