DevOps & Infra

lead21pc Bundle

Thiết kế, audit, sửa, rút gọn hoặc kiểm thử ChatGPT Custom Instructions dựa trên regression hành vi quan sát được và kiến trúc điều khiển kiểu v8.3. Dùng cho CI hành vi; không dùng cho CI/CD, chỉnh văn phong thông thường, hoặc tự áp v8.3 lên mọi tác vụ.

comedianhhh Skill

Maintain the application pipeline in career/tracker.md — new applications, recruiter screens, OAs, interview loops, offers, rejections, follow-ups — from emails, screenshots, or chat, and run the daily new-postings scan through jobs-mcp using career/targets.yaml. Use when the user types /offer, reports an interview invite, rejection, or offer, asks what to follow up on, wants the pipeline summarised, or asks to scan for new jobs.

ldphuong-vn Skill

Stage 0 of the validation pipeline — framing. Use to turn a vague idea/product/decision into a testable problem statement, a 5W1H frame, and a risk-ranked assumption list, before any gate runs. Triggers when the orchestrator routes to FRAMING or when a user idea is still fuzzy and needs structuring. Always run this first so downstream stages have something concrete to test.

ldphuong-vn Skill

Stage 1 of the validation pipeline — the hard gate. A cheap, binary, no-LLM-judgment screen that kills most ideas early before any expensive analysis. Use right after framing. Triggers when the orchestrator routes to HARD_GATE. Always run for new ideas/features; any single FAIL stops the pipeline.

matthewtaormina Bundle

Knowledgebase of current AI API providers and models (chat/agent, image, video, music/speech), documentation links, pricing, privacy/retention, and routers/gateways. Use when selecting models or vendors, comparing OpenAI, Anthropic, Google, Mistral, DeepSeek, xAI/SpaceXAI, Groq, Together, Fireworks, Vertex, Bedrock, Azure, OpenRouter, EUrouter, LiteLLM, Portkey, Sora, Veo, FLUX, Lyria, Kling, Runway, or when the user asks which model to use for chat, agents, image, video, or music.

ldphuong-vn Skill

Stage 2 of the validation pipeline — problem validation. Use to verify the problem is real with evidence rather than opinion, using The Mom Test discipline. Triggers when the orchestrator routes to PROBLEM_VAL (MUST_RUN for greenfield/pivot; ASSERT when a live product already proves demand). Always run before validating any solution.

ldphuong-vn Skill

Stage 3 of the validation pipeline — solution validation. Use to test whether the proposed solution actually moves the needle, via the Riskiest Assumption Test, Value Proposition Canvas, and an MVP/MLP plan. Triggers when the orchestrator routes to SOLUTION_VAL. Always run before market sizing — a great market for a solution nobody adopts is worthless.

ldphuong-vn Skill

Stage 6 of the validation pipeline — the decision gate. Use this whenever an idea/product/feature/decision has cleared the upstream stages and needs a Go/No-go/Pivot/Park verdict with a confidence score. Triggers when the user asks to "make the call", "should we build/ship/kill this", run a decision review, or when the orchestrator routes to DECISION_GATE. This is the stage that applies Six Thinking Hats, a weighted decision matrix, pre-mortem/inversion/red-team critique, and RAPID roles, then writes verdict + confidence + decision journal into the Dossier. Always use this rather than improvising a verdict, so the decision is evidence-based, reversible-aware, and recorded.

ldphuong-vn Skill

Stage 8 of the validation pipeline — lifecycle operations and the re-validation loop. Use to define funnel metrics, cohort retention, expand/sunset criteria, and the tripwires that trigger re-running earlier gates. Triggers when the orchestrator routes to LIFECYCLE. Use for live-product audits (case B) and after any GO to close the loop from launch to ongoing operation.

ruslankhis Bundle

Select, implement or review Google ADK workflow control flow, event consumption, session-state handoffs and callback boundaries. Use when choosing between ordinary code, sequential, parallel, bounded-loop, graph or dynamic orchestration, or repairing those runtime contracts in an existing agent. Do not activate for unrelated UI edits, general LLM explanations, standalone cloud deployment, or implementing RAG, memory or SQL services.

ruslankhis Bundle

Implement, adapt or review safe external API tools in Python and ADK projects, including selective retries, cooperative deadlines, idempotent writes and confirmation. Use for unreliable API calls, ambiguous write outcomes, unsafe tool replay or setup failures blocking ADK API verification. Do not activate for model-only tuning, general cloud deployment, UI-only changes or database query optimisation without an external-call safety problem.

ruslankhis Bundle

Build, adapt or audit a Python ADK agent's testing strategy: deterministic tool and Runner tests, live evaluation sets, simulated conversations, conformance replay and strict result checks. Use for agent behaviour, regression coverage or evaluation reliability. Do not activate for ordinary non-agent unit tests, standalone model benchmarks, deployment, or cloud cleanup alone.

ruslankhis Bundle

Build, adapt, test or troubleshoot Google ADK SQL agents using reviewed templates, selective schemas and controlled read-only execution. Use for natural-language analytics, SQL routing, generated-query reliability, or the agent's BigQuery/Gemini identity, setup and cleanup. Do not activate for ordinary SQL tuning, general database administration, RAG-only work or unrelated cloud hosting.

ruslankhis Bundle

Implement or review authenticated Google ADK tools with verified user/session ownership, keyless GCP access, Secret Manager and delegated OAuth lifecycle controls. Use for tool credential selection, cross-user access bugs, consent/refresh/disconnect flows or credential exposure in agent events. Exclude generic sign-in UI, general IAM administration, model-key billing troubleshooting and content moderation without a tool-authentication boundary.

ruslankhis Bundle

Implement, adapt or audit invocation limits, token budgets, repeated-tool protection, human approval and application cost controls in Python Google ADK agents. Use for runaway agent loops, cross-turn budget enforcement, risky actions awaiting review, or staged cost shutdown. Do not activate for a single HTTP retry policy, deployment alone, prompt-only editing, or general cloud billing administration.

ruslankhis Bundle

Select, implement, adapt or audit deployment of Python Google ADK agents on Cloud Run, Agent Runtime or GKE, including identity, packaging, browser validation, recovery and owned-resource cleanup. Use for ADK hosting decisions and deployment lifecycle work. Do not activate for general Kubernetes administration, non-ADK applications, prompt-only changes, or RAG and memory design without a deployment task.

ruslankhis Bundle

Inspect, implement or review sensitive-data boundaries in Python Google ADK agents using Sensitive Data Protection (SDP), Model Armor, strict tool contracts and guarded response release. Use for PII or credential exposure in prompts, tool calls, session history, logs or streamed answers, and for planning these controls and their cloud prerequisites. Do not activate for generic ADK setup, ordinary authentication work without a data-boundary concern, Cloud Armor WAF configuration, unrelated infrastructure cleanup, or non-agent document redaction.

ruslankhis Bundle

Diagnose, implement or review Python Google ADK optimisations on Cloud Run, Agent Runtime and GKE involving model/tool work, history, streaming completion, session handling, startup, concurrency or workload scaling. Use for measured changes or actionable optimisation plans. Do not activate for hosting selection or first deployment alone, general Kubernetes administration, standalone RAG or memory architecture, non-ADK performance work, or unrelated prompt and UI edits.

kwhi6693-web Bundle

AI image generation with OpenAI GPT Image 2, Azure OpenAI, Google, OpenRouter, DashScope, Z.AI GLM-Image, MiniMax, Jimeng, Seedream, Replicate and Agnes APIs. Supports text-to-image, reference images, aspect ratios, and batch generation from saved prompt files. Sequential by default; use batch parallel generation when the user already has multiple prompts or wants stable multi-image throughput. Use when user asks to generate, create, or draw images.

qualcomm-linux Bundle

Fetch and analyze LAVA test job logs, metadata, results, and definitions from a Foundries/Linaro LAVA instance (default lava.infra.foundries.io). Use when given a LAVA job URL or job ID (e.g. .../scheduler/job/256592) and asked to investigate a test failure, root-cause a job, or read the serial console / kernel boot log. The LAVA web UI is behind Anubis anti-bot protection so WebFetch returns 403 — use the REST API endpoints below instead.

scagogogo Skill

Check boolean properties of version numbers — type checks (IsBeta, IsStable, IsRC, etc.) and comparison predicates (IsNewerThan, IsOlderThan, IsBetween). Use for CI/CD conditionals, shell script exit-code checks, and programmatic version filtering.

scagogogo Skill

Compose the versions CLI in shell pipelines, scripts, and CI/CD — output formats, exit codes, input methods, and command patterns.

zime-skills Bundle

Builds a new Claude skill from a plain-language request, or tightens an existing one. Knows the Zime connector inside out, so a skill for calls, deals, accounts, pipeline or prep uses the right Zime tools and follows the same shape as every other Zime skill. Works for any other connector too, from a short profile of its tools. Use for "make me a skill that…", "create a skill for…", "turn this prompt into a skill", "build a skill on top of Zime", "write a skill for the HubSpot connector", "improve this skill", "why isn't my skill triggering", or "package this as a .skill". Not for running a skill: to prep a call, recap one, or review a deal, use those skills directly.

zime-skills Skill

Everything about the commercial side: one account, one deal, or the whole pipeline, as a CRM record or as a grounded analysis. Use for "pull up the Acme account", "what stage is the Acme expansion", "who owns this deal", "how do I win the Acme deal", "why is this deal at risk", "what's at risk this quarter", "what's stalled", "review Priya's pipeline", or "what should I focus on this week". Works out scope (account, deal, or portfolio) and depth (facts or judgment) from the question itself. Grounded in the live CRM and the calls behind it, never generic sales advice, and reports a dimension as unknown rather than filling it with best practice. For a company not yet in CRM, use account-research instead.

ivelten Skill

Set up or detect a Haskell Dev Container using the pre-built ivelten/haskell-devcontainer Docker image. Covers detection of an existing devcontainer setup, creating .devcontainer/ files from scratch, VS Code extension list, direnv integration, and what the image does and does not ship. Use when bootstrapping a Haskell project in a container, checking whether a devcontainer is already configured, or setting up VS Code Dev Containers for Haskell.

ivelten Skill

Set up a native (non-container) Haskell toolchain on the host with mise + GHCup. Covers the mise.toml pin (ghcup, cabal, ormolu), GHC and HLS via GHCup, the HLS/GHC compatibility rule, cabal-installed dev tools, PATH and shim wiring for direnv and editors, macOS prerequisites, and the CI mirror of the same pins. Use when bootstrapping a Haskell project without Docker, when the devcontainer is too slow on macOS, or when asked to install or pin GHC, cabal, HLS, ormolu or hlint locally.

roboco-io Bundle

Draw.io (.drawio) XML 형식의 다이어그램 생성. 아키텍처 다이어그램, 시퀀스 다이어그램, 플로우차트, ERD 등을 요청할 때 사용. AWS 아이콘을 포함한 클라우드 아키텍처 다이어그램 지원.

roboco-io Bundle

AWS Well-Architected Security Pillar review. Checks IAM, encryption, network security, and detective controls in IaC code. Integrates with owasp-review for application code.

roboco-io Bundle

AWS Well-Architected Performance Efficiency Pillar review. Checks resource selection, scaling, caching, and monitoring configurations in IaC code.

roboco-io Bundle

AWS Well-Architected Reliability Pillar review. Checks high availability, fault tolerance, backup, and disaster recovery configurations in IaC code.

roboco-io Bundle

AWS Well-Architected Framework based IaC review orchestrator. Use when users request infrastructure review, cloud architecture assessment, or Well-Architected review. Supports Terraform, CloudFormation, CDK, and Pulumi.

roboco-io Bundle

AWS Well-Architected Sustainability Pillar review. Checks energy efficiency, resource optimization, and sustainable practices in IaC code.

roboco-io Bundle

AWS Well-Architected Operational Excellence Pillar review. Checks monitoring, logging, deployment automation, and continuous improvement practices in IaC code.

antsteyer Skill

Chain the three PR passes on the PRs awaiting my review, in one go: build the comprehension brief (pr-brief), draw the flow when the mechanism is spread out (pr-flow), then leave a pending review (review-requested-prs) that reads both as context. Use when the user says "/pr-review-pipeline", "brief, flow et review", "prépare et review les PRs", or wants the three passes chained on one or several PRs. Pick this over review-requested-prs whenever the brief (and flow) must be built before the review; review-requested-prs alone is the review pass without that preparation.

guysilvere Skill

Création de visuels 3D procéduraux Three.js à partir d'une image de référence (héros produits, objets animés, dioramas) pour landings et SaaS Agence Bulles. Wrapper léger du pipeline img2threejs — reconstruction par code (primitives + shaders procéduraux), pas de meshes téléchargés. Charger quand un visuel 3D est requis.

guysilvere Bundle

Déploiement et exploitation d'une PWA Agence Bulles — RUNBOOK, déploiement Coolify + Cloudflare (proxy orange, tunnels Zero Trust), sauvegardes Cloudflare R2 (backup/restore), compression AVIF/WebP, docker-compose, staging. Charger (via ops-quality) dès qu'un déploiement, une sauvegarde, une restauration ou une mise en production est en jeu (Phase 8 du workflow).