Saga (Warp)
WarpOrchestrates spec-driven development of medium-to-large features through parallel worker agents, staged validation and a final user acceptance check.
By agent · Claude Code
Skills in the open SKILL.md format that Claude Code reads. Each one has a plain-language summary and the command to install it.
$ npx skills add <author>/<skill>Replace the placeholder with the author and skill name from the skill's page. Start a new Claude Code session afterwards so the skill is picked up.
Orchestrates spec-driven development of medium-to-large features through parallel worker agents, staged validation and a final user acceptance check.
Routes internet research through Agent Reach across 16 platforms, using CLI, API and browser backends for search, social, jobs, code, video and finance.
Drafts Warp feature documentation from product and technical specs, verifies claims in the codebase, and produces an MDX page with a review-ready draft PR.
Produces shareable HTML readouts from conversation findings or codebase research, with source references and an automatically refreshed local index.
Diagnoses and reduces Neon Postgres egress by finding high-transfer queries and fixing overfetching, unbounded results and repeated data.
Provides branch-aware S3-compatible object storage for Neon projects, with buckets, uploads, downloads, presigned URLs and standard AWS tooling.
Deploys long-running Node.js HTTP functions on Neon branches for APIs, streaming, WebSockets, webhooks and scheduled or storage-triggered workloads.
Guides developers through Google Cloud account authentication, project selection or creation, billing setup and deployment verification.
Explains how to authenticate and authorise Google Cloud users, services and external workloads with secure credential and IAM choices.
Connects Neon projects to multiple LLM providers through one branch-scoped credential, supporting SDK calls, model switching and agent workflows.
Guides FastAPI Python backend development with typed routes, Pydantic validation, asynchronous operations and maintainable API structure.
Evaluates Google Cloud workloads against Well-Architected security principles and recommends controls for identity, networks, data, operations and compliance.
Generates Google Cloud cost guidance using the Well-Architected Framework, covering visibility, resource efficiency, governance and continuous optimisation.
Generates Google Cloud reliability guidance for workload architecture, covering availability, resilience, observability and disaster recovery.
Sends brief, anonymous, unstructured Slack complaints about frustrating agent tooling or the experience of agent work.
Reports concrete, generalisable friction in agent tooling and workflows to an internal suggestion box with structured, safe context.
Runs Dart analysis, applies mechanical fixes and formats changed code using the Dart CLI or available MCP tools.
Creates and organises Dart and Flutter unit tests with package:test, including grouped cases, async checks, mocks and test commands.
Fixes Dart analysis and runtime failures by tracing errors, applying code changes, and checking results with analysis, tests and hot reload.
Resolves Dart package version conflicts by auditing constraints, upgrading compatible dependencies, and refreshing only the affected lockfile entry.
Applies Dart 3 pattern matching to validate data, destructure collections and records, parse responses and handle sealed types exhaustively.
Generates Mockito test doubles for Dart dependencies and guides stub, verification and test execution with build_runner.
Generates LCOV test coverage reports for Dart and Flutter projects using the coverage package and Dart VM tooling.
Guides developers through Dart primary constructor syntax, migration, concise constructors and common initialisation errors.
Questions
A folder with a SKILL.md file: instructions, and sometimes scripts and reference files, that Claude Code loads when a task calls for them.
Each skill page has an install command. Run it in a terminal, then start a new Claude Code session so the skill is picked up.
An instructions file is loaded in every session and describes the project. A skill is loaded only when a task calls for it, so it can be long and specific without costing context on every turn.