Code / Plan / API usage
Claude CodeRepository-aware coding with explicit plans, permissions and review.
Use the permission system for hard boundaries; a best-effort hook is not a sandbox.
[ 03 — 29 source-linked tool profiles ]
What it does, when it fits, where it fails and how to run a useful first test. Coding, research, local inference, automation, retrieval, evaluation and media—without invented ratings.
29 profiles · documentation-reviewed · no paid ranking or invented ratings
Code / Plan / API usage
Claude CodeRepository-aware coding with explicit plans, permissions and review.
Use the permission system for hard boundaries; a best-effort hook is not a sandbox.
Code / Plan-dependent
CursorEditor, cloud agents and persistent Projects in one development workflow.
Choose where tools execute before asking a coordinator to delegate work.
Code / Plan / API usage
OpenAI CodexCoding agents with sandbox policies, approvals and inspectable traces.
Cached search, indexed search and unrestricted network access are different trust decisions.
Agents / API / plan-dependent
Google AI StudioExplore Gemini modalities and turn a tested prompt into an API workflow.
Flash, Live and TTS share a family name, not one interchangeable request format.
Research / Plan-dependent
Gemini NotebookSource-grounded research and study workflows, with explicit rollout caveats.
A source-grounded answer can still inherit a bad source. Curate before you summarize.
Local AI / Local + cloud options
OllamaRun local models or connect supported cloud models behind a familiar workflow.
Check where the selected model runs. Installing Ollama does not make a cloud model local.
Automation / Cloud / self-hosted
n8nConnect models to deterministic business workflows and approval steps.
Let a model classify or draft; let explicit workflow logic decide where the output goes.
Evaluation / Cloud / self-hosted
LangfuseTrace model calls, prompts, cost and evaluation outcomes together.
Instrument the backend. Tracing secrets never belong in a browser bundle.
Code / Plan-dependent
GitHub CopilotAI assistance and agent workflows close to issues and pull requests.
Repository instructions improve context; branch protection still decides what ships.
Code / Software + inference
AiderTerminal pair programming built around your Git repository.
A small, relevant edit context often beats attaching the entire repository.
Code / Plan + deployment
v0From visual brief to a connected, deployable web application.
A working preview is the beginning of QA, not the end of development.
Code / Plan-dependent
LovableConversational app building with code sync and security review surfaces.
Exportability helps, but only a permission test proves that one user cannot read another's records.
Code / Plan + usage
Replit AgentBuild, inspect and deploy applications in a hosted development workspace.
A checkpoint is useful for development rollback; it is not automatically a backup strategy for production data.
Code / Plan / API usage
Grok BuildA coding-agent surface for the current Grok model family.
Evaluate the full harness and model combination, not just the release's benchmark table.
Research / Plan / API usage
Perplexity / SonarWeb-backed answers and research APIs with citations to inspect.
A link proves where a claim points, not that the linked page supports it.
Local AI / Local software
LM StudioA local-model desktop workflow with API and SDK integration.
Local chat and a network-exposed API need different security checks.
Local AI / Self-hosted / services
Open WebUIA self-hosted interface for local and compatible model endpoints.
Offline is an end-to-end property: models, embeddings, tools, downloads and auth all matter.
Automation / Cloud / self-hosted
DifyVisual workflows and chatflows with models, retrieval and branching.
Pick a Workflow for a bounded job; pick a Chatflow when conversation state is part of the product.
Automation / Cloud / self-hosted
FlowiseVisual assistants, retrieval flows and multi-step agent orchestration.
Use the simplest builder that expresses the job; more nodes are not evidence of better reasoning.
Agents / Library + infrastructure
LangGraphA runtime for durable, stateful agent workflows.
Persistence lets a workflow resume. Idempotency prevents its resumed steps from charging twice.
Retrieval / Library / services
LlamaIndexRetrieval and evaluation building blocks for document-aware applications.
Measure retrieval separately from answer quality so you know which component failed.
Retrieval / Cloud / self-hosted
QdrantVector retrieval with metadata filters and hybrid query pipelines.
Semantic similarity misses exact identifiers; combine lexical evidence when the task demands it.
Retrieval / Extension + database
pgvectorVector similarity search alongside relational data in PostgreSQL.
Start with exact search on a small corpus before paying the complexity cost of ANN tuning.
Evaluation / Self-hosted / services
Arize PhoenixOpenTelemetry-based tracing, datasets and evaluation experiments.
Compare variants on the same cases, and inspect what the evaluator actually saw.
Evaluation / CLI / platform options
promptfooVersioned prompt and model evaluations that can run in CI.
An eval suite earns trust by catching regressions you care about, not by producing a green percentage.
Agents / Gateway / platform options
LiteLLMProvider routing, virtual keys and budget controls behind a shared gateway.
A configured budget is not an enforced budget when the required database is missing.
Audio / Plan / usage
ElevenLabsSpeech generation, transcription and voice-agent building blocks.
Measure numbers, names and interruptions—not only how pleasant a voice sounds.
Image / Subscription
MidjourneyVisual exploration with the current V8.2 model and Edit workflow.
Check visibility settings before uploading a private reference image.
Video / Plan / usage
RunwayMedia-generation workflows with task APIs and configurable model routing.
Budget usable seconds after retries, not only the price of one generation.
Profiles are based on the linked product documentation, not fabricated hands-on scores. Recommendations count only submitted browser votes and are not identity-verified. Always check the provider’s current plan, data policy and license before adoption.