Automatic Video De-identification MCP: Reversible Masking, Restore Original — NoonAI DIS

The core workflow has three overlapping anonymization entry points (encrypt_image, encrypt_video, submit_encrypt_job) whose relationship requires careful reading to disentangle, and the job-output trio (get_job_status, get_job_result, download_job_result) can be confused. Descriptions do clarify the progression, but the boundaries between the 'one-shot' encrypt_* tools and the split upload+submit flow are not obvious from names alone.

Mostly consistent snake_case verb_noun pattern (get_*, upload_*, submit_encrypt_job, encrypt_image/video, download_job_result). Minor deviations like purchase_credits_with_card and the dual upload_file/upload_file_content suffix add slight irregularity but remain readable and predictable.

15 tools is within a reasonable range for the stated scope, and the anonymization lifecycle (upload, submit, status, result, download) is well covered. However, six of the fifteen tools are billing-related, which is heavy relative to the core de-identification purpose and makes the surface feel slightly lopsided.

The end-to-end job lifecycle (two upload paths, submission, polling, metadata, download) plus billing management is largely complete, with face detection intentionally folded into the encrypt calls. Minor gaps exist: no job cancellation/deletion and no dedicated list-jobs beyond usage summary, but core workflows have no dead ends.