ProofFetch Lite (proof_fetch, read_clean, url_watch)

Three tools (proof_fetch, read_clean, url_watch) all fetch a public URL, which creates genuine overlap risk. However, the descriptions sharply delineate them by output: full proof receipt + markdown, title + markdown only, and hash/change-detection respectively. sec_section and site_crawl are clearly distinct, so misselection is unlikely but not impossible.

All five names use snake_case, giving a consistent visual style. The pattern is not strictly verb_noun though: proof_fetch and url_watch are noun_verb, read_clean is verb_adjective, and sec_section/site_crawl are noun_noun. Readable and predictable enough, but the ordering convention wavers.

Five tools is well-scoped for a fetch/proof service, and each covers a distinct mode (provenance fetch, light read, monitoring, SEC extraction, crawling). No tool feels redundant or like padding.

The surface covers fetch-with-proof, light read, change detection, SEC section extraction, and site crawling, forming a coherent lifecycle for content retrieval and verification. Minor gaps exist (e.g., no content-diff/summary tool beyond raw hashes, no batch/multi-URL fetch), but agents can work around these.