调试
定位并修复问题，分析错误
re-frame2-pair-retro
day8
Retrospect on a re-frame2-pair session and turn it into prioritised improvement ideas for the pair skill, preload runtime, MCP surface, or upstream re-frame2 Tool-Pair contract; optionally drafts a GitHub issue the user can file. Not for ordinary re-frame2-pair operation, nor for the code/spec/framework work the body's routing matrix sends elsewhere. Requires evidence: a concrete re-frame2-pair session in this conversation (turns where the user attached, dispatched, walked traces/epochs, hot-swapped, or time-travelled), or a user-supplied recap — vocabulary alone ("retro", "what went wrong", "any improvements?") does not justify activation. Two triggers: (a) explicit pull — "retro on this pair session", "review my re-frame2-pair session", "draft an issue about that"; or (b) post-error inside a pair session — after a stack trace, a pair tool returning {:ok? false :reason :error-kind}, or an :rf.error/ or :rf.epoch/restore- trace during live pair work.
re-frame2-pair
day8
Pair-program against a running re-frame2 application via its Tool-Pair contract — attach to a live shadow-cljs nREPL, inspect a frame's app-db, dispatch events, hot-swap handlers, read the trace stream and per-frame epoch history, time-travel with restore-epoch, and replay a retained epoch strictly with replay-epoch. Use when the user is operating on (or wants to operate on) a live runtime they have running locally. Do not use for static spec reading, architecture questions, design discussion, or ordinary source edits when no runtime is involved — those belong to skills/re-frame2/ (authoring) or direct spec reading. See references/vocabulary.md for the surface glossary; vocabulary matches alone do not justify activation.
rust-development
thrashr888
Rust development workflow with quality gates, testing, and iteration patterns. Use when developing Rust code, running tests, or iterating on Rust projects.
rust-best-practices
thrashr888
Rust coding best practices for idiomatic, efficient, and maintainable code. Use when writing Rust code, reviewing code, or learning Rust patterns.
twd-tester
BRIKEV
TWD test runner agent — runs in-browser tests via twd-relay, reads failures, fixes issues, and re-runs until green. Use when you want to execute and validate existing TWD tests.
plain-bug
dropseed
Submit a bug report for the Plain framework. Use when you encounter a bug, error, or unexpected behavior. Collects context and creates a GitHub issue.
plain-upgrade
dropseed
Upgrades Plain packages and applies required migration changes. Use when updating to newer package versions.
effect-best-practices
forcedotcom
Enforces Effect-TS patterns for services, errors, layers, and atoms. Use when writing code with Effect.Service, Schema.TaggedError, Layer composition, or effect-atom React components.
vscode-window-messages
forcedotcom
Guidelines for using vscode.window.show*Message methods. Use when working with showInformationMessage, showWarningMessage, showErrorMessage.
verification
forcedotcom
Verification steps for code changes. Use after ANY code change to ensure quality, or when creating plans because those should include verification steps.
ts4023-effect-errors
forcedotcom
Fix TS4023 errors when exporting Effect-based functions. Use when TypeScript reports "has or is using name 'X' from external module but cannot be named" for Effect error types, or when knip flags error type exports as unused.
gus-cli
forcedotcom
Interact with the Gus Salesforce org via CLI (sf). Query, create, update work items; query team epics. Use when user mentions Gus, work items, epics, or GUS/Git2Gus workflows.
research-debug
iamseungpil
Research-driven debugging workflow that combines web research for similar cases, deep code analysis with task-planner-analyzer, and iterative fixes with modular-code-architect and code-reviewer. Use when encountering training collapse, gibberish generation, or architecture-level bugs where the problem may be documented in research literature. Parallel execution of research and fixes for efficiency.
sentry-issue-resolver
Arctuition
"Analyze and resolve Sentry issues by fetching detailed issue information, performing deep root cause analysis, and providing actionable solutions. Use when the user asks to: (1) Analyze a Sentry issue, (2) Debug or investigate a Sentry error, (3) Fix a Sentry issue, (4) Get root cause analysis for application errors, (5) Resolve Sentry alerts. Works with Sentry URLs to fetch stack traces, error context, and event data."
golang
cosmix
Go language expertise for writing idiomatic, production-quality Go code. Use for Go development, concurrency patterns, error handling, testing, and module management. Triggers: go, golang, goroutine, channel, interface, struct, pointer, slice, map, defer, context, error, gin, echo, fiber, cobra, viper, gorm, sqlx, go mod, go test, effective go, errgroup, sync, mutex, waitgroup.
circuit-breaker-pattern
aj-geddes
Implement circuit breaker patterns for fault tolerance, automatic failure detection, and fallback mechanisms. Use when calling external services, handling cascading failures, or implementing resilience patterns.
event-driven
cosmix
Event-driven architecture patterns including message queues, pub/sub, event sourcing, CQRS, and sagas. Use when implementing async messaging, distributed transactions, event stores, command query separation, domain events, integration events, data streaming, choreography, orchestration, or integrating with RabbitMQ, Kafka, Apache Pulsar, AWS SQS, AWS SNS, NATS, event buses, or message brokers.
rust
cosmix
Rust language expertise for writing safe, performant, production-quality Rust code. Primary language for the Loom project. Use for Rust development, ownership patterns, error handling, async/await, cargo management, CLI tools, and serialization. Triggers: rust, cargo, rustc, ownership, borrowing, lifetime, trait, impl, struct, enum, Result, Option, async, await, tokio, serde, clap, thiserror, anyhow, Arc, Mutex, RwLock, RefCell, Box, Rc, Vec, HashMap, HashSet, String, derive, macro.
browser-debugging
aj-geddes
Debug client-side issues using browser developer tools. Identify JavaScript errors, styling issues, and performance problems in the browser.
alert-management
aj-geddes
Implement comprehensive alert management with PagerDuty, escalation policies, and incident coordination. Use when setting up alerting systems, managing on-call schedules, or coordinating incident response.
before-after
cosmix
Generates before/after verification pairs for loom plans. Proves a stage actually changed system behavior by capturing state before and after implementation. Use for delta-proof verification — proving new commands, endpoints, modules, or bug fixes work by comparing system state.
receiving-code-review
guanyang
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation
test-driven-development
guanyang
Use when implementing any feature or bugfix, before writing implementation code
k8s-incident
rohitg00
Respond to Kubernetes incidents with runbooks and diagnostics. Use for outages, pod failures, node issues, network problems, and emergency response.