Skills

What it isThe control loop around a language model: prompting, tool dispatch, state and retries.

Used forRunning agents predictably in production without depending on a third-party framework.

What it isAn open protocol for exposing tools and data to AI assistants through a standard server.

Used forLetting clients such as Claude Desktop or Cursor call backend tools and data.

What it isOne tool definition shared by an in-app agent and by external MCP clients.

Used forAvoiding duplicate integrations and keeping tool behavior identical everywhere.

What it isA Python framework for building MCP servers and clients with little boilerplate.

Used forTurning Python functions into typed MCP tools, resources and prompts.

What it isAnthropic's API and official SDKs for the Claude family of models.

Used forBuilding chat, tool-using agents and structured generation on Claude.

What it isOpenAI's API and official SDKs for its GPT and reasoning models.

Used forText generation, function calling, embeddings and structured outputs.

What it isA model emitting a structured call to a named function instead of free text.

Used forConnecting models to APIs, databases and actions with typed arguments.

What it isConstraining model output to a JSON schema generated from Pydantic models.

Used forGetting machine-readable, validated results that code can rely on.

What it isRetrieving relevant documents and passing them to the model with the question.

Used forGrounding answers in private or current data and reducing hallucination.

What it isFinding items by meaning, comparing embedding vectors rather than keywords.

Used forDocument retrieval for RAG, similarity search and recommendations.

What it isA Stanford framework that treats prompts as programs to compose and optimize.

Used forBuilding LLM pipelines whose prompts are tuned automatically against metrics.

What it isOpen-weight language models from Meta (LLaMA 3) and Microsoft (Phi-3).

Used forRunning capable models on your own infrastructure for privacy or cost.

What it isIsolating what each agent step can see and do inside its own scope.

Used forStopping side effects or untrusted content from leaking into an agent's context.

What it isGuards that halt an agent run when its trace shows loops, errors or overspend.

Used forContaining runaway agent behavior before it costs money or causes damage.

What it isReusing a stable prompt prefix across calls, and tracking prompt changes like code.

Used forCutting latency and token cost, and making prompt changes reviewable and reversible.

What it isMTCNN detects and aligns faces; FaceNet turns a face into a comparable embedding.

Used forFace detection and verification pipelines in computer vision.

What it isAn open-source tool for tracing, evaluating and debugging LLM applications.

Used forInspecting every model and tool call, latency and cost in one trace view.

What it isAn open-source time-series database and monitoring system that scrapes metrics.

Used forCollecting service and cluster metrics and firing alerts on thresholds.

What it isAn open-source dashboarding tool for metrics, logs and traces.

Used forVisualizing Prometheus and other data sources and sharing live dashboards.

What it isAdding your own spans, metrics and logs to code, beyond what libraries emit.

Used forMeasuring the business-specific steps that generic monitoring cannot see.

What it isAn OpenTelemetry-based convention for describing LLM and agent spans.

Used forMaking AI traces portable across observability tools such as Phoenix.

What it isThe vendor-neutral standard unit of a distributed trace: one timed operation.

Used forFollowing a request across services to find where time and errors go.

What it isTime to first token: how long a model takes to start responding.

Used forTuning perceived speed of streaming AI features and finding slow steps.

What it isFixed test suites that score model or prompt behavior automatically.

Used forCatching quality regressions before a prompt or model change ships.

What it isCounting tokens and spend per customer, user, team and model.

Used forBilling, chargeback and spotting which tenants or features drive cost.

What it isLimits and policies on how much each tenant or feature may spend on models.

Used forKeeping AI costs predictable with caps, alerts and enforced quotas.

What it isA general-purpose language with the largest ecosystem for AI and data work.

Used forAPIs, data pipelines, ML tooling and glue code across the stack.

What it isA systems language with memory safety guaranteed at compile time, no GC.

Used forLow-latency services, compilers and performance-critical libraries.

What it isThe standard language for querying and changing relational databases.

Used forReporting, transformations and application data access.

What it isA declarative query language for property graphs, created for Neo4j.

Used forExpressing multi-hop relationship queries in graph databases.

What it isJavaScript with a static type system, compiled to plain JavaScript.

Used forLarge frontend and Node.js codebases where types catch errors early.

What it isA modern async Python web framework built on type hints and the ASGI standard.

Used forHigh-throughput REST APIs with automatic validation and OpenAPI docs.

What it isAn ergonomic Rust web framework from the Tokio team.

Used forBuilding fast, type-safe HTTP services in Rust.

What it isAn async Rust SQL toolkit that checks queries against the database at compile time.

Used forType-safe database access from Rust services without an ORM.

What it isThe most widely used asynchronous runtime for Rust.

Used forRunning many concurrent network tasks efficiently on few threads.

What it isA Python library for data validation and settings built on type hints.

Used forValidating API payloads, configs and model outputs into typed objects.

What it isPython's most widely used SQL toolkit and object-relational mapper.

Used forDatabase access from Python with composable queries or mapped models.

What it isThe database migration tool for SQLAlchemy.

Used forVersioning schema changes and applying them safely across environments.

What it isRust bindings for Python: write native Python extensions in Rust.

Used forMoving hot paths into Rust while keeping a Python interface.

What it isHTTP APIs organized around resources and standard methods.

Used forThe common contract between frontends, services and third parties.

What it isNon-blocking input and output, so one worker serves many waiting requests.

Used forHandling high concurrency in network services without extra threads.

What it isReshaping queries, indexes and access patterns so the database does less work.

Used forCutting API latency and database load, often by orders of magnitude.

What it isKeeping recent or expensive results close at hand to avoid recomputing them.

Used forServing hot reads fast and protecting databases and model APIs from load.

What it isPaging through results with a pointer to the last item rather than an offset.

Used forStable, fast pagination over large or frequently changing tables.

What it isCoordinating HTTP callbacks that systems send when events happen.

Used forReliable event-driven integrations with retries, ordering and idempotency.

What it isBreaking text into tokens, then parsing them into an abstract syntax tree.

Used forBuilding compilers, formula engines and domain-specific languages.

What it isA model of explicit states and allowed transitions held in memory.

Used forTracking complex workflows, such as lineage, deterministically and fast.

What it isTracing where each output column came from and how it was transformed.

Used forImpact analysis, debugging data issues and audit or compliance work.

What it isAI coding agents that read a codebase, edit files and run commands.

Used forFixing bugs and shipping features faster, with the engineer reviewing.

What it isAn open-source relational database; GIN indexes power its full-text search.

Used forTransactional data, fast text search and pooled connections at scale.

What it isAn in-memory graph database compatible with Cypher.

Used forReal-time graph analytics and streaming relationship queries.

What it isThe most widely used native graph database, queried with Cypher.

Used forModelling and querying highly connected data such as networks and lineage.

What it isMicrosoft SQL Server, Microsoft's relational database.

Used forEnterprise transactional systems and data warehouses.

What it isA standard columnar in-memory format shared across languages.

Used forMoving data between tools and processes without copying or converting.

What it isA fast DataFrame library written in Rust, using the Apache Arrow columnar memory format.

Used forAnalytical queries over large data with lazy, streaming execution.

What it isMicrosoft Azure's managed service for building ETL and data pipelines.

Used forScheduling and orchestrating data movement between sources and warehouses.

What it isExtract, transform, load: moving data from sources into an analytics store.

Used forFeeding warehouses and reports with clean, consistent data.

What it isSlowly changing dimensions: overwrite in place with no history (Type 1), or add a row per change (Type 2).

Used forTracking how reference data such as customers changes over time.

What it isAgreed meanings for data, and formal agreements on its schema and quality.

Used forKeeping business definitions and technical models aligned between teams.

What it isA universal desktop database client and SQL editor.

Used forExploring schemas, writing queries and managing many databases in one tool.

What it isThe standard container orchestrator; DOKS is DigitalOcean's managed version.

Used forDeploying, scaling and self-healing containerised services.

What it isInfrastructure as code: cloud resources declared in files and applied by plan.

Used forRepeatable, reviewable provisioning of clusters, networks and VMs.

What it isPackages of templated Kubernetes manifests, managed by Helm.

Used forInstalling and upgrading applications on Kubernetes consistently.

What it isMulti-stage builds compile in one image and ship only the result in another.

Used forSmaller, faster-to-build and more secure container images.

What it isRunning and maintaining virtual machines; Droplets are DigitalOcean's VMs.

Used forHosting services, build agents and tools that need a full machine.

What it isMicrosoft's platform for repositories, pipelines and release management.

Used forAutomating build, test and deployment for every change.

What it isCI agents that run on your own Kubernetes pods or VMs instead of shared hosts.

Used forFaster builds with access to private networks and custom tooling.

What it isCloudflare's DNS, global edge network, TLS certificates and web firewall.

Used forFast, secure routing for sites and APIs with DDoS and attack protection.

What it isA desktop IDE for viewing and managing Kubernetes clusters.

Used forInspecting workloads, logs and resources across clusters visually.

What it isGoogle Cloud Platform, Google's public cloud.

Used forCompute, storage, data and AI services on Google infrastructure.

What it isAmazon Web Services, the largest public cloud provider.

Used forCompute, storage, databases and managed services at any scale.

What it isAn audit standard covering security, availability, processing integrity, confidentiality and privacy.

Used forShowing customers, with evidence, that their data is handled safely.

What it isA tool that scans Python dependencies for known vulnerabilities.

Used forCatching vulnerable packages in CI before they ship.

What it isAn open-source scanner for vulnerabilities in images, code and IaC.

Used forScanning container images and configs as part of the pipeline.

What it isAuth0 is a hosted identity platform; SSO means one login across apps.

Used forSecure sign-in with enterprise identity providers and social logins.

What it isThe standard protocol for delegated authorization using access tokens.

Used forLetting apps act for a user without ever handling their password.

What it isRole-based access control: permissions granted to roles, roles to users.

Used forLeast-privilege access that stays manageable as teams grow.

What it isAn append-only record of who did what, and when.

Used forInvestigations, accountability and passing compliance audits.

What it isA static type checker for Python; strict mode allows no untyped code.

Used forCatching type errors before runtime in large Python codebases.

What it isAn extremely fast Python linter and formatter written in Rust.

Used forConsistent style and early bug detection in seconds, even on big repos.

What it isPython's most widely used testing framework.

Used forUnit and integration tests with fixtures and readable assertions.

What it isThe operating system behind most servers, and its standard shell.

Used forOperating servers, scripting automation and debugging in production.

What it isThe standard distributed version control system.

Used forTracking changes, branching and collaborating through code review.