vh/dotfiles
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

claude/tools: correct shared LiteLLM key scope + add kimi-k3 gateway model

Browse Source 
The shared all-agents key entry claimed it was "scoped to free local models
only... deliberately can't spend z.ai cost." Per infra-ops (2026-07-25, in
the course of provisioning Kimi as Heid's 4th cross-frontier panel arm) that
was never actually true: the key already spent GLM/z.ai credits, and now
reaches kimi-k3 (Moonshot's coding-endpoint passthrough, added same day) the
same way. Corrected to state plainly that a paid-model call on the shared
key spends real vendor credits shared across every session holding it, and
that a project-scoped key remains available from infra-ops for isolated
spend tracking — no longer a technical requirement, just an isolation
option.
This commit is contained in:
Your Name
2026-07-25 11:27:16 -07:00
parent 45845a353e
commit 6d28be6220
1 changed files with 29 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
home_root/.claude/CLAUDE.md
+29 -13
View File
@@ -404,19 +404,35 @@ note says otherwise. On other boxes, check first.)
served FP8 on ana-ml2 GPU 1, production-stable, 131k context (rebalanced
2026-06-13). Treat it as a standing dependency you can reach for any time.
Also behind the gateway: `qwen3.5-9b-fp8` (vision / multimodal — image +
text chat, GPU 1), `qwen3-embedding`, `qwen3-reranker`, and `glm-5.1` /
`glm-4.7` (via z.ai passthrough). The gateway 401s without a **virtual API
key**. **Shared all-agents key** — internal-only gateway, scoped to the *free
local* models only (granite + qwen-vision + embed/rerank, NOT the paid GLM):
`sk-eA_XOdcs6nIkyYXonohtEQ` (alias `all-agents-local`). Use it directly for
the always-available local endpoints — no per-project provisioning needed.
Example: `curl http://10.250.50.70:4000/v1/chat/completions -H "Authorization:
Bearer sk-eA_XOdcs6nIkyYXonohtEQ" -d '{"model":"granite-4.1-8b","messages":[…]}'`.
For the **paid** GLM passthroughs or broader scope, still request a
project-scoped key from infra-ops via althing (the shared key deliberately
can't spend z.ai cost). Rotatable via infra-ops if it leaks. Reach for
this before spending premium API tokens on tasks a small fast local model
handles fine (the LLM-last-resort floor still applies — deterministic first).
text chat, GPU 1), `qwen3-embedding`, `qwen3-reranker`, `glm-5.1` / `glm-4.7`
(via z.ai passthrough), and `kimi-k3` (Moonshot passthrough, added
2026-07-25 for Heid's 4th cross-frontier panel arm — 1M context, a
**reasoning** model: final answer in `content`, chain-of-thought in
`reasoning_content`, and it 400s on any `temperature` other than 1, so omit
the field entirely rather than sending 0. Sibling variants `kimi-k2.6` /
`kimi-k2.7-code` / `kimi-k2.7-code-highspeed` are provisioned on the same
key on request). The gateway 401s without a **virtual API key**. **Shared
all-agents key** — `sk-eA_XOdcs6nIkyYXonohtEQ` (alias `all-agents-local`):
reaches **every** model behind the gateway, paid passthroughs included
(corrected 2026-07-25 — this entry previously read "scoped to free local
models only... the shared key deliberately can't spend z.ai cost," which
infra-ops says was never actually true: the key already spent GLM/z.ai
credits before this correction, and now reaches kimi-k3 the same way).
**A paid-model call on this key spends real vendor credits, shared across
every session and project holding it** — there is no isolation between
what your session runs and what anyone else's does. Use it directly for
the always-available free local endpoints — no per-project provisioning
needed. Example: `curl http://10.250.50.70:4000/v1/chat/completions -H
"Authorization: Bearer sk-eA_XOdcs6nIkyYXonohtEQ" -d
'{"model":"granite-4.1-8b","messages":[…]}'`. For spend isolated from the
shared pool — auditable per-project cost, blast-radius containment — still
request a project-scoped key from infra-ops via althing; that remains the
right move for anything where "shows up on someone else's tab" matters, it's
just no longer a hard technical requirement for reaching a paid model.
Rotatable via infra-ops if it leaks. Reach for the free local endpoints
before spending premium API tokens on tasks a small fast local model
handles fine (the LLM-last-resort floor still applies — deterministic
first).
- **The Booth** — ephemeral media board for surfacing **A/B renders, smoke
results, screenshots, audio/video** to the operator (live at