aria_skills/litellm/SKILL.md
Manage LiteLLM proxy, models, and API spend tracking.
npx skillsauth add najia-afk/aria_moltbot aria-litellmInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Manage LiteLLM proxy, query available models, track API spend and provider balances.
exec python3 /app/skills/run_skill.py litellm <function> '<json_args>'
List all available models.
exec python3 /app/skills/run_skill.py litellm litellm_models '{}'
Check proxy health.
exec python3 /app/skills/run_skill.py litellm litellm_health '{}'
Get spend logs.
exec python3 /app/skills/run_skill.py litellm litellm_spend '{"limit": 20}'
Get total spend summary.
exec python3 /app/skills/run_skill.py litellm litellm_global_spend '{}'
Get wallet balances from Kimi and OpenRouter.
exec python3 /app/skills/run_skill.py litellm provider_balances '{}'
Returns:
{
"kimi": {"provider": "Moonshot/Kimi", "status": "ok", "available": 17.72, "cash": 0, "voucher": 17.72},
"openrouter": {"provider": "OpenRouter", "status": "ok", "limit": 10, "usage": 0.3, "remaining": 9.7}
}
GET /litellm/models - List modelsGET /litellm/health - Health checkGET /litellm/spend - Spend logsGET /litellm/global-spend - Global summaryGET /providers/balances - Provider wallets| Provider | Env Variable | Currency |
|----------|-------------|----------|
| Kimi/Moonshot | MOONSHOT_KIMI_KEY | USD |
| OpenRouter | OPEN_ROUTER_KEY | USD |
| Local (MLX) | N/A | Free |
tools
# aria-working-memory Persistent short-term memory operations for Aria runtime context. ## Purpose - Store, retrieve, rank, checkpoint, and prune working memory. - Provide high-relevance context for decision and response flows. ## Main Tools - `remember` - `recall` - `get_context` - `checkpoint` - `restore_checkpoint` - `forget` - `update` - `reflect` - `sync_to_files`
development
```skill --- name: aria-unified-search description: "🔎 RRF-merged search across semantic + graph + memory backends" metadata: {"aria": {"emoji": "🔎"}} --- # aria-unified-search Unified search engine that queries semantic memories (pgvector cosine), knowledge graph (ILIKE), and traditional memories (text match), then merges results via Reciprocal Rank Fusion (RRF) with content-hash deduplication. ## Architecture ``` Query ↓ (parallel to 3 backends) ├── SemanticBackend (pgvector cosi
tools
# aria-telegram Telegram Bot API integration skill. ## Purpose - Send and reply to messages through bot credentials. - Fetch updates and basic bot metadata. ## Main Tools - `send_message` - `reply_to_message` - `get_updates` - `get_me`
tools
# aria-sprint-manager Sprint planning and board flow management skill. ## Purpose - Operate goal board as Product Owner. - Move, prioritize, and report sprint progress. ## Main Tools - `sprint_status` - `sprint_plan` - `sprint_move_goal` - `sprint_report` - `sprint_prioritize`