Analyze existing agent configuration and context files in a repository — .cursor/rules, AGENTS.md, CLAUDE.md, .claude/ settings, MCP configs, and other agent-facing context. Assesses quality, completeness, cross-tool consistency, and identifies conflicts or gaps. Use when running an insight scan's agent context analysis phase, or when you want to audit how well a repository's agent configuration supports coding agents.

Analyze agent conversation logs from Cursor, Claude Code, or other AI coding tools to identify patterns that affect agent performance. Discovers log locations automatically, parses conversations for repeated failures, frustration signals, excessive iteration, and tool misuse. Use when running an insight scan's agent log analysis phase, or when you want to understand what agent conversation history reveals about performance bottlenecks in a repository.

Produce a structured inventory of every markdown and configuration file that coding agents consume in a repository — entry points (AGENTS.md, CLAUDE.md), always-on rules (.cursor/rules), hooks, skills, and MCP servers. Classifies each file by which agent harness reads it (Claude / Cursor / cross-agent) and rates its usefulness. Use when running an insight scan's context inventory phase, or when you want a filterable tree view of the context agents see.

Use when someone asks to review, proofread, or lint documentation for errors, typos, or quality issues. Checks spelling, grammar, header level consistency, placeholder documentation in examples, and ensures all docs have a Description and at least one example unless clearly a reference page. Useful for documentation quality checks, markdown review, or checking for typos.

Expert guidance for configuring and deploying the OpenTelemetry Collector. Use when setting up a Collector pipeline, configuring receivers, exporters, or processors, deploying a Collector to Kubernetes or Docker, or forwarding telemetry to Dash0. Triggers on requests involving collector, pipeline, OTLP receiver, exporter, or Dash0 collector setup.

Configures trace spans, defines custom metrics, sets up log exporters, and optimizes sampling strategies for OpenTelemetry instrumentation. Use when instrumenting applications with traces, metrics, or logs. Triggers on requests for observability, telemetry, tracing, metrics collection, logging integration, or OTel setup.

OpenTelemetry Transformation Language (OTTL) expert. Use when writing or debugging OTTL expressions for any OpenTelemetry Collector component that supports OTTL (processors, connectors, receivers, exporters). Triggers on tasks involving telemetry transformation, filtering, attribute manipulation, data redaction, sampling policies, routing, or Collector configuration. Covers syntax, contexts, functions, error handling, and performance.

Develop or review put.io SDK repositories, API clients, and client libraries across TypeScript, Swift, Kotlin, and similar packages. Use when adding or changing namespaces, tightening request or error types, aligning SDK behavior with backend and app usage, updating SDK verification flows, or checking how an SDK repo should be documented and released.

Set up put.io frontend repos: README/CONTRIBUTING/SECURITY, CI, package scripts, verify commands, release workflows, deploy pipelines, test harnesses, and publish/deploy flows. Use for repo setup, repo cleanup, project setup, configuring CI or deployment, or making a package/app/SDK repo documented, verifiable, and deliverable. Skip feature code, SDK API work, Vite+ migrations, and self-verification.

Validates emotions, maintains empathetic tone, avoids premature reassurance, and prevents repetitive or hollow responses during extended emotional exchanges. Use when the person is expressing grief, pain, fear, hopelessness, anger, or other intense emotions across multiple turns — especially when the conversation extends into prolonged emotional territory. Addresses LLM behavioral failure modes (strategy preference collapse, premature escalation, grief-flattening), not clinical intervention. Not a substitute for professional mental health support.

Produce premier client-ready proposals, capability decks, and case studies for Metis Strategy BD work. Use whenever the user asks to create, build, draft, rebuild, or iterate on a proposal, capabilities deck, case study, BD follow-up, or pitch document. Trigger on: "proposal," "premier proposal," "capabilities deck," "capabilities piece," "case study," "BD follow-up," "build the proposal," or "proposal" with BD context (transcript, prospect, champion, buyer, SVP, CDO, CIO). Also trigger when source material (transcript, PPTX, notes) is provided and a polished client-ready output is requested. Output is a 16:9 landscape PDF (13.33" x 7.5", 10-25 pages) with full-bleed photography, branded graphic devices, and ReportLab layout. Do NOT use for: PowerPoint decks (metis-pptx); whitepapers (metis-whitepaper); one-pagers or internal reports (metis-pdf-creator); SOWs/MSAs (metis-legal-drafting); capability maps (metis-capability-map).

Creates Postgres INSERT triggers that fire Edge Functions for document chunking and embedding generation. Configures the ingestion pipeline from raw document insert to vector storage. Use when building RAG ingestion, embed-on-insert pipelines, database-driven document ingestion, or automated embedding workflows.

Executes EXPLAIN ANALYZE via MCP to debug slow queries, identify missing indexes, detect sequential scans, and optimize query plans. Use when debugging slow SQL, analyzing query plans, finding unused indexes, optimizing Postgres queries, or investigating index-not-used issues.

Implements server-side auth session management with HTTP-only cookie synchronization for Next.js, SvelteKit, and other SSR frameworks. Depends on PKCE auth flow. Use when implementing SSR authentication, server-side session sync, HTTP-only auth cookies, Next.js Supabase auth, or SvelteKit Supabase auth.

Implements, evaluates, and deploys multi-armed bandit algorithms — including Thompson Sampling, UCB, epsilon-greedy, LinUCB, EXP3, and contextual bandits. Covers algorithm selection, experiment harnesses, offline evaluation (IPS, Doubly Robust), infrastructure patterns, and correctness verification. Use when the user asks about multi-armed bandits, exploration-exploitation tradeoffs, adaptive experiments, A/B testing alternatives, online optimization, bandit-based recommendation or personalization systems, or contextual bandits.