134 skills · sorted by relevance
Organize human review in Arize by defining label choices, numeric ranges, or free-text feedback fields. Manage review queues and reviewers, submit annotations for individual records, and apply existing labels to project spans, or individual processing records, in bulk.
Inspect existing LLM application records in Arize to investigate errors, slow operations, or behavior changes. Export traces, which group related operations, or individual operations and conversation sessions as JSON using the ax command-line tool.
Investigates distributed application performance using PostHog APM (OpenTelemetry span) data via MCP. Use when the user asks about service traces, slow HTTP/database spans, error spans, trace IDs, or span attributes — not AI observability traces or product logs. Uses posthog:query-apm-spans, posthog:apm-trace-get, posthog:apm-services-list, posthog:apm-attributes-list, and posthog:apm-attribute-values-list.
The English summary is being prepared. Showing the original description.
INVOKE THIS SKILL when creating, managing, or using annotation configs on Arize (categorical, continuous, freeform), or applying human annotations to project spans via the Python SDK. Configs are the label schema for human feedback on spans and other surfaces in the Arize UI. Triggers: annotation config, label schema, human feedback schema, bulk annotate spans, update_annotations.
The English summary is being prepared. Showing the original description.
Inspect how a task ran across agents by collecting recorded execution spans, or timed processing steps. Build a trace tree showing agent activity, duration, errors, and bottlenecks, with a summary of the execution flow.
Create and run evaluations in Arize using an LLM as the judge. Define scoring criteria, connect them to actual data fields, evaluate recorded operations or experiment results, and set up continuous scoring for new data.
Investigate slow or failing requests in Google Cloud using Cloud Trace. Filter traces by duration or error status, inspect spans representing individual operations, and retrieve related Cloud Logging entries to investigate bottlenecks and exceptions.
Configure OpenTelemetry to send AI application traces, records of execution steps, to Confident AI. Reuse existing exporters, choose the regional endpoint, and add AI-specific attributes without requiring the deepeval Python package.
Add OpenTelemetry tracing to MATLAB code. Use when the user asks to "add tracing", "instrument with spans", "add OpenTelemetry", "trace my code", "add observability" (when about tracing), or mentions "spans", "distributed tracing", or "OTel tracing" in the context of MATLAB functions. Covers span creation, parent-child context propagation, error handling, attributes, events, and semantic conventions.
The English summary is being prepared. Showing the original description.
Finds secrets, tokens, passwords, and API keys that can leak through generated serialization: Python dataclass repr and asdict, attrs, pydantic model_dump, NamedTuple, JavaScript JSON.stringify and util.inspect, structured logs, tracing spans, and error reports. Use when asked to "check for secret serialization", "credential in repr", "repr=False audit", "secret in spans", "token in logs", or when a change adds a credential field to a dataclass, model, or config object, or adds code that stringifies whole objects or every kwarg into telemetry.
The English summary is being prepared. Showing the original description.
Bridge layer connecting AlgebraicJulia/Catlab.jl to plurigrid/asi. Wires ACSets (attributed C-sets), wiring diagrams, decorated cospans, and the AlgebraicJulia ecosystem (AlgebraicDynamics, AlgebraicPetri, AlgebraicRewriting, Decapodes) into the ASI skill graph. ACSets generalize relational databases with categorical semantics; every diagram, network, and model in the ecosystem is an ACSet.
The English summary is being prepared. Showing the original description.
Track the full Toulmin epistemic ancestry of every claim produced by a multi-agent swarm. Each agent output is tagged with an argument chain (claim / data / warrant / relationship) derived from SwarmTracer spans and LineageEdge records, so operators can zoom into any sub-claim and see precisely which agents asserted it, what triggered each assertion, and how contradictions propagated or were synthesised away. The SwarmTracer pattern — wrapping the SwarmAgentExecutor to capture spans, record messages, and build a lineage graph — is the concrete runtime substrate. Argumentative lineage lifts that graph into the Toulmin vocabulary so reasoning provenance is both machine-queryable and human-interpretable.
The English summary is being prepared. Showing the original description.
INVOKE THIS SKILL for LLM-as-judge evaluation workflows on Arize: creating/updating evaluators, running evaluations on spans or experiments, tasks, trigger-run, column mapping, and continuous monitoring. Use when the user says: create an evaluator, LLM judge, hallucination/faithfulness/correctness/relevance, run eval, score my spans or experiment, ax tasks, trigger-run, trigger eval, column mapping, continuous monitoring, query filter for evals, evaluator version, or improve an evaluator prompt.
The English summary is being prepared. Showing the original description.
Choose which related skill to use and in what order for planning, building, debugging, or maintaining software. It explains routes through the available workflows and helps select the next step based on your situation and project size.
Turn a large engineering objective into a written plan with steps sized for individual pull requests, or proposed code changes. Each step includes context, dependencies, checks, and completion criteria so another agent can continue the work.
Turn an unclear feature idea into a structured specification with goals, boundaries, and testable success criteria. Split broader requests into independent modules when needed, then move through planning, task breakdown, and implementation with human review.
Investigate slow Paperclip runs, page requests, and CI suites using Honeycomb traces, which show timed operations. Find relevant traces, separate overlapping work, identify dominant delays, and configure optional tracing or add missing measurements.
Review multi-step forms to find information users must enter more than once. Suggest automatic filling, reuse options, and preserving data between steps, while explaining relevant accessibility requirements and exceptions for security or validity.
Answer questions about past decisions, usual practices, and changes over time by reasoning across Hindsight memories. When information spans multiple memory banks, combine the answers and identify their sources, while acknowledging gaps or contradictions.
Build a tabbed project status page with goals, workstreams, owners, risks, and next actions. Publish it privately on claude.ai, then refresh the same URL from current sources and report what changed since the previous update.
Plan and coordinate authorized penetration tests across multiple stages, from reconnaissance to access, privilege escalation, movement between systems, and impact assessment. Assign specialist skills, revise blocked paths, and maintain evidence and progress records for the final report.
Route reverse engineering and security tasks to a specialist skill based on the target, purpose, and available tools. Establish the case scope and authorization status, then prepare supported tools before handing off the work.
Investigate LLM application failures with Phoenix CLI by reviewing execution records, recording observations, and grouping mistakes into counted categories. Inspect datasets and experiments to identify which failure types deserve evaluation tests first.
Set up Phoenix tracing for LLM applications in Python or TypeScript. The skill guides automatic and custom recording of individual operations, conversation grouping, metadata, feedback, and production settings such as batching and personal data masking.