d4rken
UserA multiplexing load-balancer proxy for LLM providers.
Categories
Indexed Skills (6)
fork-workflow
How ClankerMux work is branched, merged and released. Read this when starting a branch, merging into main, cutting a version bump, pulling a fix from the tombii/better-ccflare upstream, or merging an inbound PR from an external contributor.
git-recovery
Recovering uncommitted work in the live ClankerMux checkout, and the worktree recipes for reading or checking out another branch. Read this if the main checkout is on an unexpected branch, if WIP appears missing, or when you need to inspect another branch without moving HEAD.
rate-limiting
Rate limiting, usage windows, cooldowns and recovery in ClankerMux. Read this before touching 429/529 handling, cooldowns, account selection or routing, usage polling, the usage cache, rate-limit display in the dashboard, or the accounts.rate_limit_* columns.
running-clankermux
Building, starting and exercising ClankerMux locally — server commands, ports, the production Caddy topology, account management via the HTTP API, maintenance endpoints, and how to send a safe test request through the proxy.
db-migrations
Adding a column, table or migration to the ClankerMux SQLite database. Read this before editing packages/database/src/migrations.ts or adding any schema change — there is a two-step requirement whose omission fails silently on every existing live database.
qwen-provider
Working on the Qwen/DashScope provider or its streaming transform. Read this before changing tool-call handling, streaming chunk assembly, or anything under the Qwen provider — its wire format differs from standard OpenAI in a way that breaks naive implementations.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.