Add "omp" to the Kimi Code approved coding-agent whitelist

Request: whitelist the omp coding agent on api.kimi.com/coding, so omp users can use Kimi Coding membership without impersonating another client.

Context

• omp (oh-my-pi) is an open-source terminal coding agent (https://omp.sh), a fork of pi (https://pi.dev). Users want to drive Kimi Coding (kimi-for-coding / K2.7 Code) from omp using their Kimi membership.
• api.kimi.com/coding admits requests only from approved clients by inspecting User-Agent. Today omp cannot use Kimi Coding without spoofing an approved client identity (for example KimiCLI/*).
• Spoofing is prohibited by the Kimi Code Community Guidelines ("Don't spoof or alter client identity information"), so we will not ship that. We want our own identifier whitelisted instead.

Request

• Whitelist the omp/<version> User-Agent (and X-Msh-Platform: omp) on api.kimi.com/coding, on the same terms as Kimi CLI, Claude Code, Roo Code, and Kilo Code. omp currently sends omp/17.3.7.
• omp sends its own stable identifier. This is not a request to spoof any other client.

Why this is small

• The whitelist is prefix-based (KimiCLI/ and claude-code/ both pass today), so a single allowlist entry covers omp.
• Endpoint, model id, and the OAuth device flow are unchanged; omp already implements the public device-code flow against auth.kimi.com.

Precedent

• Add "pi" to the Kimi Code approved coding-agent whitelist #2185 (pi)
• [Feature Request] Add Cline to the Kimi Code whitelist of supported coding agents kimi-cli#2322 (Cline)
• Add Zoo Code to the third-party coding agent API whitelist kimi-cli#2416 (Zoo Code)

References

• Community Guidelines: https://www.kimi.com/code/docs/en/kimi-code/community-guidelines.html
• Third-party agents: https://www.kimi.com/help/kimi-code/third-party-agents

Start by reviewing the User-Agent allowlist for api.kimi.com/coding and the precedent issues linked in the request. Add omp/ version handling and X-Msh-Platform: omp on the same terms as the existing approved clients, then verify omp can use the endpoint without spoofing another client identity.