Agent Skills
› tracewayapp/traceway
› traceway-learn
traceway-learn
GitHubTraceway可观测性平台使用指南，解释产品各功能模块、术语含义及界面导航。用于回答用户关于功能查询、概念解释及工作流设置的问题，不涉及数据读写或代码配置。
Trigger Scenarios
询问Traceway某功能或页面如何工作
查找特定功能位置
解释平台术语含义
Install
npx skills add tracewayapp/traceway --skill traceway-learn -g -y
SKILL.md
Frontmatter
{
"name": "traceway-learn",
"description": "The full guide to Traceway, the observability platform. Explains every part of the product and how it is meant to be used: issues and exception grouping, endpoints and the impact score, tasks, spans and traces, logs, metrics and dashboards, sessions, AI traces, profiles, monitors, incidents, status pages, post-mortems, alert rules and channels, on-call teams, schedules, overrides, escalation policies and pages, contact methods, roles, organization views, settings and agent access. Use when the user asks how a Traceway page or feature works, where to find something, what a term means, or how to set up a workflow such as paging or a status page."
}
Learn Traceway
Traceway is an observability platform. Applications send telemetry: OpenTelemetry over OTLP/HTTP from backends, and the Traceway SDKs from browser and mobile apps. Traceway turns it into issues, request and task analytics, traces, logs, metrics and dashboards, session replay, AI cost tracking and profiles. On top of that it runs uptime monitors, public status pages, alerting and on-call paging.
This guide explains every part of the product: what each page shows, the terms it uses, how teams are meant to use it, and how an agent reads or changes it through the traceway CLI or the MCP server.
How to use this guide
- Answer "how does X work", "where do I find X" and "what does X mean" from the topic files below. Read only the one or two the question needs.
- Quote the exact names the dashboard uses (tab names, buttons, rule types), so the user can find them.
- When the user wants live data, switch to the
tracewayskill (or the MCP tools): it holds the debugging and performance playbooks. - When the user wants something changed, the
traceway-writeskill covers the changes agents are allowed to make. Everything else is done in the dashboard, and the topic files say so. - When the user wants an application instrumented, use the
traceway-setupskill.
Core ideas
- Organization owns people, projects, dashboards, on-call and status pages. Project is one application, and almost every page is scoped to it. A trace can cross several projects of one organization.
- Roles:
owner,admin,user,readonly, with per-project overrides foruserandreadonlymembers. - Time windows: every telemetry page is bounded by a range kept in the URL (
?preset=24h, or?from=&to=). Presets: 5m, 30m, 60m, 3h, 6h, 12h, 24h, 3d, 7d, 1M, 3M. CLI and MCP reads take--sinceor--from/--to. - Spans are the primitive. Endpoints, tasks, AI traces and issues are spans promoted at ingest. One trace id links a request to its downstream calls, logs and exceptions.
- Alert, page, incident are different things. An alert is a rule sending a notification. A page is an on-call escalation that keeps going until someone acknowledges it. An incident is a period when a monitor was down (or one recorded by hand), with an optional public story on a status page.
Topics
| Topic | File | MCP resource | Covers |
|---|---|---|---|
| Issues | issues.md | traceway://knowledge/learn/issues |
Exceptions, fingerprinting, the issue and occurrence pages, archiving, regressions |
| Endpoints and tasks | endpoints-and-tasks.md | traceway://knowledge/learn/endpoints-and-tasks |
P50, P95, P99, impact score, Apdex, expected-performance marks, requests, background tasks |
| Spans, traces and logs | traces-and-logs.md | traceway://knowledge/learn/traces-and-logs |
The span explorer, whole-trace waterfalls, distributed traces, log search, live tail |
| Metrics and dashboards | metrics-and-dashboards.md | traceway://knowledge/learn/metrics-and-dashboards |
Metric names and tags, aggregations and rate, dashboards, widgets, templates, starring |
| Sessions, AI traces, profiles | sessions-and-ai.md | traceway://knowledge/learn/sessions-and-ai |
Session replay, LLM calls, conversations, flagged content, flame graphs |
| Monitors and status pages | monitors.md | traceway://knowledge/learn/monitors |
HTTP, TCP and browser checks, runs, incidents, public status pages, post-mortems |
| Alerts | alerts.md | traceway://knowledge/learn/alerts |
Every rule type and its settings, channels, cooldown, snooze, history |
| On-call | on-call.md | traceway://knowledge/learn/on-call |
Teams, schedules, layers, overrides, escalation policies, pages, contact methods |
| Organization and access | organization.md | traceway://knowledge/learn/organization |
Projects, roles, the home page, organization views, Connection, Account, Settings, agent access |
Page map
| Page | Path | Topic |
|---|---|---|
| Dashboard (home) | / |
organization.md |
| Issues | /issues, /issues/<hash> |
issues.md |
| Logs | /logs |
traces-and-logs.md |
| Endpoints | /endpoints, /endpoints/<endpoint> |
endpoints-and-tasks.md |
| Tasks | /tasks |
endpoints-and-tasks.md |
| Spans | /spans, /spans/<traceId> |
traces-and-logs.md |
| Profiles | /profiles |
sessions-and-ai.md |
| Sessions | /sessions (browser and mobile projects) |
sessions-and-ai.md |
| AI Traces | /ai-traces |
sessions-and-ai.md |
| Dashboards | /dashboards |
metrics-and-dashboards.md |
| Monitors | /monitors (tabs Monitors, Status Pages, Post-Mortems) |
monitors.md |
| Public status page | /status/<slug> |
monitors.md |
| Alerts | /notifications (tabs Channels, Rules, History) |
alerts.md |
| On-Call | /on-call (tabs Pages, Overview, Teams, Schedules, Policies) |
on-call.md |
| Connection | /connection |
organization.md |
| Account | /account |
organization.md, on-call.md |
| Settings | /settings (owners and admins) |
organization.md |
| Organization | /organization (Servers, Issues, Monitors, On-Call, Projects) |
organization.md |
| Project setup | /setup |
organization.md |
Browser and mobile projects only show Issues, Sessions, Alerts, On-Call and Connection.
Common questions
| Question | Answer in |
|---|---|
| Why did two errors land in the same issue, or one error split in two? | issues.md, Fingerprinting |
| What does the impact score mean and why is this endpoint Critical? | endpoints-and-tasks.md |
| What happens when I acknowledge a page? Who gets paged next? | on-call.md, Pages and Escalation policies |
| Who is on call right now, and how do I swap a shift? | on-call.md, Schedules and Overrides |
| Snooze or disable a noisy alert? | alerts.md, Disable versus snooze |
| Why is my monitor down after one failure, or not down after several? | monitors.md, States and runs |
| How do I publish uptime to customers? | monitors.md, Status pages |
| Where do I write up the incident? | monitors.md, Post-mortems |
| Why is a span missing from the waterfall? | traces-and-logs.md |
| Why does my counter metric only go up? | metrics-and-dashboards.md, Aggregations |
| Can an agent change things in Traceway? | organization.md, How agents get access |
Version History
- 3c3ed3e Current 2026-10-18 17:33
