dsh-llm-kiro
Run the following command in DeepSeek Harness:
dsh plugin install caopu16/dsh-llm-kiro
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install caopu16/dsh-llm-kiro in your DeepSeek Harness setup to mount the Kiro credential adapter; the plugin source is available at https://github.com/caopu16/dsh-llm-kiro
About this plugin
If you already have a signed-in AWS Kiro (CodeWhisperer) account but want to route Claude-family models through the DeepSeek Harness alongside open-weight ones, dsh-llm-kiro bridges that gap. It mounts the Kiro provider route into Harness and reads the sign-in credentials that Kiro IDE or kiro-cli write on the same machine, with no second API key, no extra credential store, and no write-back race.
Under the hood, the adapter maps Kiro's vnd.amazon.eventstream frames into Harness text, reasoning blocks, and tool dispatch. Streaming usage is derived from the context-occupancy percentage each response reports, tool names are auto-aliased to match the wire grammar, and thinking markers are injected per effort level. For open-weight models the wire path is direct, while Claude models (authorized by egress rather than entitlement) can be pointed through an HTTP CONNECT tunnel whose interior handles TLS, so the proxy never sees request bodies or bearer tokens. A full error taxonomy covers auth failures, rate limits, and transport aborts.
It is built for developers who already live in the Kiro ecosystem and want a single Harness settings.yaml to govern both proprietary and self-hosted models. No extra key ceremony, no profile-file surgery, and an upgrade path that is just the same plugin install command run again.
Use Cases
- Reuse an existing Kiro sign-in to serve Claude models through Harness without filing for a second API key
- Pin a compliance egress tunnel for Claude while keeping open-weight models on a direct path in one settings.yaml
- Consolidate proprietary and self-hosted model routing into a single provider entry for the whole team
Best For
- Daily Kiro IDE or kiro-cli users who want Claude in their existing Harness loop
- AI engineers consolidating multi-provider model routing inside one DeepSeek Harness profile
- DevOps or platform engineers needing a compliant egress path for Claude while keeping open-weight models open
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.