1-3-Cloud-Adoption-Skills

Deploy, verify, upgrade, or roll back the anthropic_stream_guard LiteLLM plugin on a docker-compose LiteLLM gateway. Use when asked to install/uninstall the plugin, wire a LiteLLM deployment for native Claude Code clients, or verify the gateway stream fix.

Install, configure, verify, or troubleshoot a side-by-side `codex-forky` command that routes normal Codex tool/code-execution turns through an existing forky service to the MaaS execution backend such as LiteLLM/Huawei MaaS GLM, while routing non-tool ordinary turns and image turns directly to Codex ChatGPT/OAuth. Assumes forky is already installed and running, usually by claude-code-maas-hybrid-router.

Nightly (or on-demand) memory hygiene for AI coding agents — the thing native auto-memory does NOT do. Runs a "dream" pass over project memory to deduplicate entries, validate that every referenced file path still exists (so the agent never acts on "confidently wrong" stale references), compress survivors into a compact index, and detect when a new finding conflicts with an approved rule or CLAUDE.md — emitting a review-required conflict candidate and a proposed CLAUDE.md patch (never auto-applied). Use when memory has grown noisy/stale, on a nightly schedule, or whenever the user asks to dream/consolidate/clean up project memory, distill repeated workflows, or recall approved context.

Install and configure forky (github.com/vladharl/forky) end-to-end so `claude-forky` routes Claude Code planning, image, and classifier traffic through Claude OAuth while sending normal tool/code execution to Huawei Cloud MaaS (GLM-5.2) through an existing local LiteLLM proxy. Use when the user wants one Claude Code command with Claude-quality reasoning and vision plus lower-cost MaaS execution, while preserving plain `claude` for Claude.ai OAuth connectors. Sets up forky as a systemd user service, applies local compatibility patches (vision routing, Claude Code system/developer message-role normalization, and Anthropic cache-control TTL ordering), writes a `claude-forky` wrapper that uses ANTHROPIC_BASE_URL without ANTHROPIC_AUTH_TOKEN, adds copy-friendly mouse settings, trusts `/root` when requested so local permissions are honored, and merges plan-mode hooks into `~/.claude/settings.json`. Assumes LiteLLM is already running on :4000 and that the user has logged into Claude Code (`claude /login`) so OAuth c

Deploy, configure, validate, troubleshoot, or extend an OpenAI-compatible API proxy backed by PostgreSQL, Prometheus, and Grafana, routing through Huawei ModelArts MaaS (ap-southeast-1) with multi-key load balancing. TRIGGER when the task involves LiteLLM proxy deployment, Docker Compose stack with litellm_config.yaml, Huawei MaaS model routing, virtual key or budget management, Prometheus/Grafana observability for LLM traffic, custom_callbacks.py TTFT/TPOT/ITL metrics, multi-key load balancing, an optional self-hosted SearXNG search MCP (`web_search`/`fetch_url`) as an alternative to Exa, wiring a `claude-glm` (claude-code-router) client through the proxy, or any reference to `LITELLM_MASTER_KEY`, `HUAWEI_MAAS_API_KEY`, `MCP_TOKEN`, or `docker compose` with this stack.

Execute and troubleshoot Huawei Cloud server migration in this repository. Support both current batch existing-target SMS workflow (`terraform apply` + `scripts/mgc_sms_existing_target_batch.py`) and single-source SMS/rsync workflow (`scripts/mgc_migrate.py`). Use when users ask to migrate ECS/on-prem VMware, run migration in `ap-southeast-3`/`la-*`, resume after pause from `out/resume_checkpoint_latest.json`, monitor tasks in `out/task_poll_latest.json`, or diagnose SMS/MGC errors such as SMS.1404, SMS.0515, SMS.6504, SMS.6602, SMS.6603, SMS.6617, SMS.7605, SMS.7703, and SMS.8115. Also use for Chinese requests like “跨区域迁移”, “MGC/SMS 迁移流程”, “批量迁移恢复”, or “迁移排障”.

使用当前仓库的脚本执行华为云 DRS MySQL 并行迁移，覆盖环境校验、任务创建、预检查、启动同步、监控、切换前检查、切换后校验与回滚计划。

Execute and troubleshoot Huawei Cloud server migration in this repository. Support both current batch existing-target SMS workflow (`terraform apply` + `scripts/mgc_sms_existing_target_batch.py`) and single-source SMS/rsync workflow (`scripts/mgc_migrate.py`). Use when users ask to migrate ECS/on-prem VMware, run migration in `ap-southeast-3`/`la-*`, resume after pause from `out/resume_checkpoint_latest.json`, monitor tasks in `out/task_poll_latest.json`, or diagnose SMS/MGC errors such as SMS.1404, SMS.0515, SMS.6504, SMS.6602, SMS.6603, SMS.6617, SMS.7605, SMS.7703, and SMS.8115. Also use for Chinese requests like “跨区域迁移”, “MGC/SMS 迁移流程”, “批量迁移恢复”, or “迁移排障”.

Install, verify, diagnose, and harden GitHub Copilot Chat in VS Code when Huawei Cloud MaaS GLM or another Huawei MaaS OpenAI-compatible model is used through OAI Compatible Copilot. Use when Copilot stalls, emits malformed or truncated tool calls, writes partial files, shows path drift, needs Huawei MaaS model settings, needs the patched OAI Compatible Copilot VSIX installed, or needs a portable debug bundle.

Deploy Huawei CodeArts CLI on macOS and add a codearts-litellm wrapper that connects CodeArts to an OpenAI-compatible LiteLLM gateway, including model config, local auth proxy, language behavior patches, and session-scoped --yolo support. Use when installing CodeArts from scratch, configuring CodeArts with LiteLLM or MaaS through LiteLLM, fixing codearts-litellm wrappers, or reproducing this setup on another Mac.

Configure a side-by-side codex-glm command that routes Codex CLI to Huawei Cloud MaaS glm-5.1 through a CCR /v1/responses shim while preserving the original codex command.

Deploy a CSS/OpenSearch code-search MCP server on Huawei Cloud ECS from scratch, so that Claude Code (claude-glm) can search a GitHub repo's code and docs as native MCP tools. Use when the task is to provision a new Huawei Cloud CSS cluster and ECS, index a code repository into CSS, and expose it as a searchable MCP tool for AI coding agents.

Reverse engineer IBM mainframe applications and guide COBOL/JCL/CICS/VSAM/DB2 modernization into Java Spring Boot microservices, HTML/CSS/JS frontend, tests, Docker, and Kubernetes. Use when Codex is asked to analyze legacy COBOL, JCL, CICS, copybooks, assembler utilities, mainframe batch/online entry points, or to migrate monolithic mainframe systems to cloud-native Java.

Use when analyzing or migrating Databricks CDC/Delta workflow scripts into a Huawei Cloud Chile MRS + OBS + Apache Hudi demo, including synthetic CDC data generation, notebook-triggered automation, Delta-to-Hudi replacement, deployment troubleshooting, and smoke validation. Also use when continuing the dockone ExampleApp demo package under the Codex outputs folder.

Build AIOps agents for Huawei Cloud that use CSS/OpenSearch as a Splunk replacement, integrate LTS/CTS/AOM/CES for monitoring and audit, employ LangGraph state machine orchestration with approval gates, LlamaIndex knowledge retrieval, Prometheus/Grafana metrics, and auto-remediation via FunctionGraph. Use when Codex must design, provision, or generate an AIOps Agent for automatic anomaly identification, alert governance, cross-service log correlation, runbook-driven remediation, action-level enforcement (L0 read-only, L1 suggest, L2 approve+execute, L3 forbidden), or CSS-based O&M analytics on Huawei Cloud.

Deploy, operate, document, or troubleshoot the Huawei Cloud SEC EDGAR big-data lifecycle monitoring POC using OBS, MRS Spark, DWS, ECS, Superset, Terraform, PowerShell credential bootstrap, pipeline scripts, DWS loading, and the monitoring website. Use when Codex needs to reproduce this end-to-end environment, explain the scripts, continue deployment, fix failures, English/localize the websites, or create BI dashboards for the SEC EDGAR financial-topic dataset.

Use this skill when the user needs Codex to SSH into a remote Huawei Cloud ECS, or any Linux host, from a corporate network where direct SSH is blocked and access must go through an HTTP/HTTPS proxy using CONNECT. It covers checking proxy variables, proving TCP reachability, handling encrypted private keys without persisting passphrases, and executing remote commands reliably.

Build automatic context compression and tiered recall for enterprise agents that reduces token consumption while preserving critical facts, supports recall probes for compression quality verification, artifact probes for key information extraction, just-in-time retrieval that loads context only when needed, high-fidelity compaction that distills context windows, progressive disclosure from summary to detail, token cost visibility per compression operation, and Huawei Cloud deployment. Use when Codex must design, provision, or generate an agent context compression system with AI-powered summarization, tiered recall layers, compaction strategies, recall verification probes, artifact extraction probes, or token-optimized context management.

Build cross-session persistent memory for enterprise agents that auto-captures tool usage, generates semantic summaries, injects relevant context in future sessions, supports tiered retrieval with token cost visibility, skill-based search, privacy controls with sensitive content exclusion, and audit trails. Use when Codex must design, provision, or generate an enterprise agent memory system with hook-based capture, AI-powered compression, ChromaDB or CSS/OpenSearch vector retrieval, progressive disclosure, 5-layer memory architecture (L1 working, L2 episodic, L3 skill tree, L4 semantic index, L5 config), FTS5 SQLite retrieval, Markdown memory files, self-improving memory loops, Huawei Cloud deployment, or privacy-compliant agent memory.

Build project-level knowledge persistence with RAG for enterprise agents that combines learned memory (derived insights that are not re-computed) with document RAG (raw knowledge), supports GraphRAG for connected facts and relationship traversal, MCP integration for tool-accessible memory, skill-based procedural memory (how-to not just what-happened), hybrid retrieval across learned and document stores, Huawei Cloud CSS/OpenSearch and OBS deployment, privacy controls, and audit trails. Use when Codex must design, provision, or generate a project memory RAG system with learned memory storage, document RAG pipeline, GraphRAG for entity relationships, MCP tool registry for memory access, skill procedural memory, hybrid search, or enterprise knowledge persistence.

Configure, validate, document, or troubleshoot a local Docker-based Dify NL2SQL proof of concept that uses Dify workflows, an OpenAI-compatible LiteLLM model such as GLM-5.1, PostgreSQL, pgAdmin, and a safe read-only SQL execution gateway. Use when setting up Dify-to-database communication, building NL2SQL workflows, creating project reports, or adapting this architecture to another local or cloud database.

Benchmark and autoscaling test harness for Huawei Cloud CSS (OpenSearch). Use when the task is to evaluate CSS performance, run load tests, validate query quality, test data-node horizontal autoscaling, or generate consolidated benchmark reports for CSS clusters.

Build enterprise RAG knowledge agents for public-sector and regulated document collections. Use when Codex must design, provision, or generate an Enterprise RAG Agent Pack for laws, policies, official correspondence, PDFs, scans, procedures, FAQ, court/tax/social-security/healthcare/public-utility knowledge bases, multilingual Portuguese/Spanish/English cited answers, RAGFlow document parsing and OCR, LlamaIndex retrieval agents and workflows, OpenSearch/CSS hybrid retrieval, OBS storage, GaussDB/RDS metadata and audit, Huawei Cloud MaaS LLM inference, ECS demo portals for document upload and search, ACL enforcement, citation generation, evidence-chain prompts, evaluation sets, Terraform deployment, or anti-hallucination RAG workflows.

Install, configure, validate, troubleshoot, or explain RAGFlow with Huawei Cloud MaaS / ModelArts MaaS through RAGFlow's OpenAI-API-Compatible provider. Use when Codex needs to deploy RAGFlow with Docker Compose, connect it to Huawei MaaS chat models such as glm-5.1, register MaaS models in RAGFlow, verify UI/API login and LLM calls, handle Docker or RAGFlow startup issues, or protect MaaS API keys from logs and telemetry.

Use this skill when building or pitching an AI-powered customer intelligence platform for telecom operators on Huawei Cloud. TRIGGER when the user needs — AI contact center (AICC) demo architecture, telco churn prediction pipeline, ASR + LLM deployment for call analytics, executive POC strategy for telecom, data sovereignty compliance (Mexico LFPDPPP), demo design with deterministic fallback, or Huawei Cloud ECS GPU deployment for AI workloads. Also use when preparing telco executive presentations that combine live demos with strategic narrative.

Execute and troubleshoot Huawei Cloud cross-region server migration with MGC/SMS and Terraform in this repository. Use when users ask to migrate ECS across regions (especially la-north-2 to la-south-2), run the end-to-end workflow (`terraform init/apply` + `scripts/mgc_migrate.py`), validate prerequisites, map tfvars to runtime env vars, inspect migration output JSON, or diagnose SMS/MGC errors such as SMS.6602, SMS.6603, SMS.6617, SMS.7605, and SMS.8115. Also use for Chinese requests like “跨区域迁移”, “MGC/SMS 迁移流程”, or “迁移排障”.

Execute and troubleshoot Huawei Cloud server migration with Terraform in this repository. Prefer SMS first, and fallback to rsync staged migration when source is SMS-incompatible. Use when users ask to migrate ECS/on-prem VMware (especially la-north-2 and la-south-2), run the end-to-end workflow (`terraform init/apply` + `scripts/mgc_migrate.py`), validate prerequisites, map tfvars to runtime env vars, inspect migration output JSON, or diagnose SMS/MGC errors such as SMS.6504, SMS.6602, SMS.6603, SMS.6617, SMS.7605, and SMS.8115. Also use for Chinese requests like “跨区域迁移”, “MGC/SMS 迁移流程”, or “迁移排障”.

Use this skill when deploying or integrating Langfuse for LLM observability, tracing, generations, usage, latency, cost, errors, evaluations, prompts, or LiteLLM/application instrumentation. Langfuse is not a MaaS token platform, does not issue or manage MaaS API keys, and normally does not call MaaS providers directly.

Build, instrument, troubleshoot, or explain Agent applications that call Huawei Cloud MaaS through an OpenAI-compatible endpoint while using OpenLLMetry / Traceloop / OpenTelemetry for observability. Use when Codex needs to add OpenLLMetry to an Agent, LangChain/LlamaIndex/OpenAI SDK flow, LiteLLM proxy, Cline/Claude-Code-router style MaaS client, or any Huawei MaaS LLM application; generate safe examples; configure OTLP/Traceloop export; validate traces; or prevent MaaS API keys and prompts from leaking into telemetry.

Use when the user wants a generic, deployable Huawei Cloud architecture or demo for high-frequency, high-volume, complex transactions using DMS for Kafka plus GaussDB, with Java application code, idempotent processing, partitioning, retry/DLQ, and schema examples. Also use for transaction, payment, order, ledger, clearing, or risk-processing flows on Kafka + GaussDB.

Use this skill when porting SQL Server or vanilla PostgreSQL code to Huawei GaussDB (Kernel 505.x / openGauss-based). TRIGGER when source code or SQL contains any of — `Microsoft.Data.SqlClient`, `SqlBulkCopy`, `NEXT VALUE FOR`, `ON COMMIT DROP`, `CREATE TEMPORARY SEQUENCE`, `dbo.`, `NVARCHAR`, `ISNULL(`, `GETDATE()`, `OPTION (MAXDOP`, `TOP N`, `@@IDENTITY`, `WITH (NOLOCK)`, `[bracketed]` identifiers, `#tempTable`; OR csproj references `DotNetCore.GaussDB`, `DotNetCore.EntityFrameworkCore.GaussDB`, `HuaweiCloud.Driver.GaussDB`, `HuaweiCloud.EntityFrameworkCore.GaussDB`, `Npgsql.EntityFrameworkCore.PostgreSQL` alongside a GaussDB target; OR appsettings has `Persistence.Provider=GaussDb`; OR user mentions GaussDB / openGauss / 高斯数据库 / SHA256 SASL / `password_encryption_type` / apuração GaussDB migration.

Build, adapt, or explain a local finance migration demo from SQL Server to PostgreSQL through Babelfish. Use when Codex needs to create a small host-local SQL Server source database, initialize Babelfish/PostgreSQL, migrate a banking FinanceDemo workload through the TDS port, validate SQL Server/Babelfish/PostgreSQL query parity, or troubleshoot this demo's Docker, FreeTDS, sqlcmd, Babelfish initialization, md5 auth, UTF-8, or LD_LIBRARY_PATH issues.

Use this skill when migrating Cloudera/CDH Hadoop, Hive, Spark, or Impala workloads to Huawei Cloud MRS. It covers source simulation, MRS cluster provisioning, OBS data landing, Hive external table creation, Spark SQL migration, report view migration, and end-to-end parity validation. Use when the user mentions Cloudera, CDH, HDP, or any Hadoop/Spark/Hive migration to Huawei Cloud MRS, OBS, or FusionInsight.

Use this skill when setting up a financial risk control pipeline on Huawei Cloud. It helps configure OBS for raw and result data storage, MRS for Spark-based risk analysis and anomaly detection, and DWS for data warehousing and regulatory reporting. The skill covers risk scoring, AML/KYC compliance, cross-border monitoring, structuring detection, and automated report generation without relying on environment-specific details.

Build, run, or adapt a Teradata-to-Huawei-Cloud-DWS migration demo or migration toolkit, especially for finance analytics workloads. Use when Codex needs to scaffold a local Teradata-source simulation, create a minimal Huawei Cloud DWS cluster, migrate schemas/data/report SQL to DWS, validate row counts and report parity, optimize DWS tables/marts, generate migration reports, or manage DWS demo resources. Also use for real Teradata export planning through JDBC/BTEQ/TPT into DWS.

Use this skill when configuring Huawei Cloud Firewall (CFW) for financial institutions such as banks, insurance companies, or fintech platforms. Triggers include: CFW setup for banking, PCI DSS compliance firewall, financial network segmentation, IPS configuration for finance, ACL rules for banking services, intrusion prevention for payment systems, cloud firewall hardening for regulated industries, or any scenario requiring enterprise-grade firewall protection with financial regulatory compliance (PCI DSS, ISO 27001, NIST, local banking regulations).

Use this skill when building a conversational BI experience where business users ask questions in natural language and get SQL-generated answers with auto-selected visualizations. It covers text-to-SQL prompt design with schema context, multi-provider LLM fallback chains, SQL safety validation, auto-retry on query errors, follow-up question suggestions, and Streamlit dashboard patterns. This is the foundational version — use it for standard ChatBI deployments.

Use this skill when building an AI-powered document risk analysis pipeline that combines OCR text extraction, LLM-based risk scoring, and structured result storage. It covers Huawei Cloud OCR integration, serverless FunctionGraph chains, LLM prompt design for risk assessment, fallback patterns when models are unavailable, and DWS/GaussDB result storage with three-layer schema design.

Use when the user wants to prepare a local Karmada lab, install and verify the Karmada control plane, deploy the tested member1/member2 stateless failover PoC in /root/karmada, switch traffic between Kubernetes clusters, validate cutover behavior, or tear the environment down. This skill favors the repo's proven scripts and manifests over ad hoc setup so the PoC is faster and more accurate.

Use this skill when migrating Databricks tables, notebooks, SQL warehouse flows, or Spark pipelines to Huawei Cloud. It helps analyze the source workload, map Databricks patterns to OBS plus MRS Spark and Hive, generate sanitized migration scripts, validate result parity, and compare execution behavior without relying on environment-specific details.