Four tools with complete input schemas and basic descriptions, but significant gaps in parameter documentation and output schema specification. All tools share identical parameter sets (url, headers, max_length, start_index) with minimal differentiation in descriptions. No output schemas documented. Parameter descriptions are generic and lack constraints (e.g., no guidance on max_length bounds, start_index behavior, or header format). Error handling is present but not documented in tool descriptions. Tools follow verb_noun naming convention (fetch_*) which is positive, but descriptions are brief (34-48 chars) and lack context on when to use each variant or what data structures are returned.

• Document output schemas for each tool. E.g., fetch_html should return {content: string, status: number, url: string, headers: object} so agents know what fields to extract.
• Add constraints to numeric parameters: 'max_length: number (1 - 1000000, default 5000), maximum characters to return. Larger values may timeout or exhaust memory.' and 'start_index: number (≥0, default 0), character offset into the response. If ≥ content length, returns empty string.'
• Expand tool descriptions to 100 - 150 chars with context: 'Fetch a website and return the HTML content. Use this to extract raw HTML structure; use fetch_markdown for readable text, fetch_txt for plain text without markup.' Include a note on when each variant is appropriate.
• Document the 'headers' parameter security model: 'Optional HTTP headers (e.g., {"User-Agent": "MyBot/1.0"}). Do NOT pass Authorization tokens or secrets, use server-side configuration instead.' Reference the secret-injection pattern.
• Add error recovery guidance to descriptions: 'May fail if URL is invalid, unreachable, or on a private network (127.0.0.1, 192.168.*). If fetch fails, verify the URL is public and accessible.'