Build faux product UI that looks like a real product, as a still illustration or an animated demo with a cursor. Browser and app windows, dashboards, phones, notifications, terminals, bento grids and cursor walkthroughs, framed, masked and cropped the way a marketing page, feature card or docs hero needs them. Use whenever someone wants a fake or mocked UI, a product demo or walkthrough animation, a hero or feature-card illustration of an app, a product shot in the style of a polished marketing site, or an animated cursor doing things in an interface, even if they never say "faux UI". React and plain CSS.

Use when running the retrospective on a piece of work that has just finished — the retro, normally the step straight after `prep-pr` opens the pull request — to write and commit the session-metrics card for the branch, read what the session actually cost against the difficulty declared before it started, and turn that into concrete changes to the skills, `AGENTS.md`, the wiki, the tests or the brief. Also the owner of the card — nothing else writes one carrying the pull request's identity.

Debug failed or wrong-output workflow executions using executions tools. Load when the user reports execution failures, unexpected node output, empty parameter values after a successful run, or a node showing a red or failed expression error. Also load when a live run you started for the user fails with an error from a node or an external service (for example a 401), before you explain the cause to the user.

Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn/ui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for "shadcn init", "create an app with --preset", or "switch to --preset".

Convex backend specialist. Use this agent for any code inside a `convex/` directory — function definitions, schemas, indexes, queries, mutations, actions, HTTP endpoints, cron jobs, file storage, auth wiring, and component installation. Knows the object-form function syntax, validator patterns, resource limits, and component ecosystem that generic Claude routinely gets wrong.

Convex backend specialist. Use this agent for any code inside a `convex/` directory — function definitions, schemas, indexes, queries, mutations, actions, HTTP endpoints, cron jobs, file storage, auth wiring, and component installation. Knows the object-form function syntax, validator patterns, resource limits, and component ecosystem that generic Claude routinely gets wrong.

Store and query vector embeddings using Amazon S3 Vectors, a cost-effective long-term vector storage service with its own API namespace (s3vectors). Triggers on: create S3 vector bucket, vector index, store embeddings, semantic search, RAG vector storage, similarity search, vector database, migrate from other vector databases. Do NOT use for: querying tabular data (use querying-data-lake), S3 object storage, or hundreds/thousands of sustained QPS (use OpenSearch).

Create and secure S3 buckets following AWS best practices for access control, encryption, monitoring, and remediation of misconfigurations. Use when the user wants to secure a new bucket, audit an existing bucket, fix a security finding, configure encryption, or enable logging and monitoring. Do NOT use for general S3 data operations, S3 Tables setup, or discovering existing data assets.

Builds generative AI applications on Amazon Bedrock. Covers model invocation (Converse API, InvokeModel), RAG with Knowledge Bases, Bedrock Agents, Guardrails, and AgentCore. Use when invoking models, setting up Knowledge Bases, creating agents, applying guardrails, deploying to AgentCore, troubleshooting Bedrock errors (ThrottlingException, AccessDeniedException), or choosing models (Claude, Llama, Nova, Titan). ALSO USE for prompt caching setup and debugging, quota health checks and throttling diagnosis, cost attribution and tracking, migrating between Claude model generations (4.5 to 4.6 to 4.7), chunking strategies, API selection (Converse vs InvokeModel), guardrail capabilities, and model selection. NOT for custom model training, Rekognition, or Comprehend.

Builds, configures, debugs, and optimizes AWS observability using CloudWatch (Logs Insights, Metrics, Alarms, Dashboards, EMF), X-Ray, CloudTrail, and ADOT. Covers Log Insights query syntax (fields, filter, stats, parse, pattern, join, subqueries), alarm configuration (metric, composite, anomaly detection, missing data treatment), dashboard design, custom metrics (PutMetricData, EMF, metric filters), X-Ray tracing (ADOT, sampling rules, annotations vs metadata), ADOT collector config, and CloudTrail auditing. Use when the user mentions CloudWatch, Log Insights, alarms, INSUFFICIENT_DATA, dashboards, custom metrics, EMF, X-Ray, traces, sampling, CloudTrail, who deleted, ADOT, OpenTelemetry, observability, monitoring, synthetics, canaries, or troubleshooting alarm behavior. Do NOT use for application logging setup, container log drivers, or security threat detection.

AWS SDK for Python (boto3/botocore) development patterns. You MUST use this skill when writing Python code that uses AWS services via boto3 or botocore. This includes creating service clients or resources, configuring sessions and credentials, handling errors with ClientError, using paginators and waiters, S3 file transfers and presigned URLs, DynamoDB table operations, and any boto3/botocore client configuration. Use this skill whenever Python code imports boto3 or botocore, or when the user asks about AWS operations in Python.

Create managed Iceberg tables using Amazon S3 Tables (s3tables API namespace) with automatic compaction and snapshot management. Sets up table bucket, namespace, table, schema, Glue catalog registration, partitioning, IAM access control. Triggers on: create table, data lake table, analytics table, structured data storage, S3 Tables, Iceberg, Athena table, partitioning strategy, access permissions. Do NOT use for: importing files (use ingesting-into-data-lake), vector storage (use storing-and-querying-vectors), querying existing tables (use querying-data-lake), or locating existing table (use finding-data-lake-assets).

Create a production-ready, secure Amazon S3 bucket with versioning and a lifecycle policy. Use when the user asks to create an S3 bucket, enable versioning, add lifecycle rules, transition objects to Glacier or an archive tier, block public access, or encrypt a bucket. Delivers infrastructure-as-code (AWS CDK and CloudFormation) with secure defaults — Block Public Access, default encryption, a TLS-only bucket policy, versioning, and a lifecycle rule that transitions objects to Glacier after 90 days — following AWS Well-Architected best practices.

Execute an implementation plan with rigorous checklist-driven verification. Every task is pre-flighted, worked, validated with proof, and checked off. Every phase is validated as a whole. If proof is missing, the agent goes back. Honesty is non-negotiable — broken trust means the plan is not done. Triggers: 'implement this plan', 'execute the plan', 'start working on the plan', 'do the plan', 'carry out the phases', 'run the plan', 'work through the plan', 'carry out this plan', 'begin implementation'. Do NOT use for plans that have not been reviewed — use plan-review first.

Ready up Jira tickets for refinement, including brand-new not-yet-keyed backlog items, by gathering context from linked incidents or interviewing the user when no incident exists, populating a YAML template against the project's 'ready for refinement' standard, validating it against a JSON schema plus gap-detection heuristics for testability and vagueness, and generating a structured markdown document. Use when asked to ready up, prepare for refinement, draft a new backlog ticket, or write up a specific ticket key.

Investigate a candidate library, dependency, or file format against a fixed question set and end with exactly one cited recommendation (adopt / keep_current / reject / needs_more_research) -- never "it depends." Validates the structured output against a JSON schema before writing a human-readable finding. Use when choosing between candidate npm packages, dependencies, or file formats for a specific integration point, or when a plan or ADR decision hinges on an unverified factual claim about a library (does it escape by default, does it bundle cleanly, is it maintained). Do NOT use for one-line version lookups, purely subjective preference calls with no verifiable claim, or a technology choice already settled by an existing ADR -- supersede the ADR instead. Triggers: "evaluate this library", "investigate this dependency", "compare X vs Y", "properly evaluate this technology", "which format should we use", "is this package maintained".

Search Google Scholar for academic papers and author profiles. Use when discovering papers to triage — by keyword, author, or year range. Returns titles, authors, abstracts, and links. Triggers: search papers, find papers, google scholar, search scholar, find author, author profile, literature search, paper discovery, search academic literature.

Runtime type guards, narrowing, and exhaustiveness checking for TypeScript — typeof/instanceof/in guards, custom and generic type predicates (value is T), discriminated union narrowing, assertion functions (asserts value is T), branded/nominal types validated at runtime, and exhaustive switch statements with a never guard. Use when replacing an unsafe `as` assertion with a real runtime check, fixing "Object is possibly undefined", modeling impossible states out of existence, writing an `is*` predicate, or making a switch over a union fail to compile when a case is missing.

Write an original deadpan two-person political-satire interview sketch (a mock-interview bit, not a monologue, panel show, or sitcom scene) using terse questions, absolutely-confident literal-minded answers, repetition, bureaucratic logic, and escalating contradiction, in the tradition of John Clarke and Bryan Dawe's satirical news interviews. Use this when a user asks for a satirical interview, mock-interview sketch, deadpan two-hander, or "Clarke and Dawe style" bit built around a policy, institution, or current-events topic. Do NOT use this for stand-up monologue jokes, sketch comedy with more than two speaking roles, sitcom-style scenes with stage direction and physical comedy, or any request to reproduce an actual historical Clarke and Dawe transcript verbatim — this skill generates new satire, it does not retrieve or reconstruct existing scripts.

TypeScript design patterns that use the type system to enforce correctness at compile time — the builder pattern with required-field tracking, type-safe state machines and event emitters, dependency injection, plugin systems, recursive/deep-readonly utility types, type-safe module encapsulation, advanced generic constraints, and typed API clients. Use when structuring a class or module so invalid usage fails to compile rather than throwing at runtime, e.g. "how do I stop `.build()` being called before required fields are set" or "how do I type an event emitter so payloads match their event name".