febrikinali20/logscout

# LOGSCOUT `LOGSCOUT` is a 4-agent platform that ingests raw production log streams, classifies severity, detects anomalies, and writes a structured root-cause incident report — all in one MiMo function-calling loop. No regex pipelines, no alert fatigue. ## ⚙️ Pipeline raw_logs ─▶ parser ─▶ anomaly ─▶ classifier ─▶ rca ─▶ incident.json flash pro pro pro summary.md ## 🤖 Agents | # | Agent | Model | Role | |---|-------|-------|------| | 01 | `parser-agent` | `mimo-v2-flash` | Structure raw lines into `events[]` | | 02 | `anomaly-agent` | `mimo-v2.5-pro` | Statistical + semantic outlier detection | | 03 | `classifier-agent` | `mimo-v2.5-pro` | Severity + service + cause-class tagging | | 04 | `rca-agent` | `mimo-v2.5-pro` | Root-cause analysis + remediation plan | ## 🔥 Why MiMo V2.5 - **1M context** — load full pod log (250K lines) without sliding windows - **function-calling** — RCA agent invokes runbook + diff viewer in one round-trip - **structured JSON** — incident schema enforced at decode - **flash tier** — parser on cheap mimo-v2-flash, full reasoning on pro ## 💰 Token Economics | Agent | Model | Input | Output | Cost | |-------|-------|-------|--------|------| | parser | mimo-v2-flash | 16K | 4K | $0.0028 | | anomaly | mimo-v2.5-pro | 80K | 10K | $0.110 | | classifier | mimo-v2.5-pro | 120K | 14K | $0.162 | | rca | mimo-v2.5-pro | 140K | 18K | $0.194 | | **total** | | **~356K** | **~46K** | **~$0.469** | Projected: 50 incidents/day × 360K = **18M tokens/day** · ~540M tokens/month · team of 3 SREs ## 🚀 Run Locally git clone https://github.com/febrikinali20/logscout.git cd logscout python3 -m http.server 8080 ## 📦 Deploy vercel --prod ## 🔗 Links - MiMo Platform: https://platform.xiaomimimo.com/ - MiMo AI Studio: https://aistudio.xiaomimimo.com - Orbit 100T: https://100t.xiaomimimo.com/ Built on **MiMo V2.5** · Apache-2.0

标签：后端开发