This week, a usage review on Juejin quietly poked at an industry pain point: WorkBuddy, a desktop client that bundles subscriptions for a dozen-plus models including DeepSeek, Kimi, and MiniMax, denies paid users API access—they can only sit in the client window and manually copy-paste. A developer responded by writing an open-source bridge tool (middleware that connects two different protocols) called workbuddy-to-dsh, which "taps" the paid capability out of the client.
What this is
workbuddy-to-dsh is a local bridge tool consisting of three zero-dependency (no third-party libraries needed) Node scripts: 127.0.0.1:8790 exposes an OpenAI-compatible endpoint (the industry's standard protocol format for calling LLMs), 127.0.0.1:8792 serves a web console, and a native plugin installs into DeepSeek Harness (an open-source AI chat client). The console integrates usage stats, request logs, model catalog, 8 environment checks, auto check-in, and conversation testing in one screen. Logs store only metadata (time, model, token counts, etc.)—never conversation content. 31 domestic models and 26 international models can be synced into third-party clients with one click.
Industry view
This story has two sides. Supporters argue that a model subscription is essentially a user purchasing "calling rights"—the client locking down the API is selling users as traffic; since the tool only records metadata and never touches conversation content, its privacy boundaries are reasonably clean. Opposition is equally sharp: client user agreements almost certainly prohibit this kind of reverse engineering (taking software apart to see how it works); if the upstream changes its credential (login identity info) storage mechanism or encryption, the bridge can break at any moment; the more serious critique is that this tool essentially encourages "freeloading on wrappers"—users pay one subscription but extract the API value of multiple models, which will erode the aggregation app business model over the long term. We note that when aggregation clients like WorkBuddy launch, they almost always promise "cheaper than official API"—that price gap itself rests on the assumption that "users won't reverse-bridge."
Impact on regular people
For enterprise IT: After multi-model procurement, they can finally see in the console where the money goes and which model offers the best value—but with login credentials scattered across clients, compliance audits (checks on whether IT operations are compliant) leave blind spots. For working professionals: Technical users willing to tinker can upgrade models from "chat box" to "script invocation," but those without coding skills remain locked inside the client. For the consumer market: If wrapper apps frequently block bridges, they push users back to official APIs, which raises prices across the ecosystem.