菜单
Autonomous Sentry triage and fix loop. Pulls unresolved Sentry issues, investigates each in parallel, auto-dispatches up to 3 high-confidence fixes as spec loops, and resolves issues on merge. Triggers on: sentry fix, fix sentry errors, autofix sentry, sentry autofix.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Address Greptile PR review comments — read inline comments and summary, fix issues locally, push, reply to each thread, then tag Greptile to re-score and resolve threads.
End-to-end feature development loop. You describe a feature, iterate on the plan, then the team implements, reviews, ships a PR, and handles the configured PR-review bot autonomously. Notifies you at milestones. Use ONLY when the user wants the full autonomous implement→review→PR→verify loop for a multi-file feature. Do NOT use for: quick bug fixes, single-file edits, exploratory/discussion tasks, or anything the user wants to drive step-by-step.
Fetches and debugs a Langfuse trace OR session by ID or URL. A `.../sessions/<id>` URL renders a turn-by-turn conversation transcript (USER → TOOLS → ANNIE) across all the session's traces. A trace renders a span-tree overview with auto-suggested drill targets, then drills into a chosen section to surface system prompts, tool calls, and tool results. Renders both Anthropic block-style and langchain `tool_calls`-key messages, and strips noise (tool-call ids — incl. the giant gemini `__thought__` ids — plus `additional_kwargs`/`response_metadata`/message ids). Supports comparing two spans side-by-side. Use when the user provides a Langfuse trace/session ID or URL, mentions a langfuse.* link, or asks to debug, inspect, diagnose, or compare an LLM trace or conversation. Defaults to anyformat credentials (LANGFUSE_TRACING_*) and host (langfuse.anyformat.ai); other hosts work if LANGFUSE_HOST and LANGFUSE_PUBLIC_KEY/LANGFUSE_SECRET_KEY are set.
Query and debug anyformat's Datadog observability (logs, metrics, traces, monitors, incidents) via the plugin:datadog:mcp server, accounting for our non-standard loguru log structure. Use when searching or debugging anyformat logs/metrics/traces/monitors/incidents in Datadog, or when a Datadog log search for anyformat-core/backend returns 0 results.
Commit changes grouped by logical chunks, push to a feature branch, link a Linear ticket, and create a labelled PR. Targets 'dev' if it exists on origin (anyformat convention), otherwise the repo's default branch. Triggers on: commit, push, create PR, ship it, send PR.
Address failing CI checks on the current PR — group failures by root cause, fix locally, push, then watch the re-run. Pairs with react-to-greptile (comments) for full PR clean-up.
| name | sentry-fix |
| description | Autonomous Sentry triage and fix loop. Pulls unresolved Sentry issues, investigates each in parallel, auto-dispatches up to 3 high-confidence fixes as spec loops, and resolves issues on merge. Triggers on: sentry fix, fix sentry errors, autofix sentry, sentry autofix. |
| argument-hint | [project1,project2] [dry-run] | single <sentry-id|sentry-url|gh-issue-url|plan-path> |
| allowed-tools | ["Read","Glob","Grep","Bash","Agent","Skill","Edit","Write","Task","ScheduleWakeup","ToolSearch"] |
Autonomous loop: investigate unresolved Sentry issues in parallel, dispatch the top 3 as spec --auto-approve loops, resolve on merge. Keeps #tech informative, not noisy.
User invokes /sentry-fix
│
▼
┌─ PREREQ ────────────────┐ → reference/prereqs.md
│ Load env vars │
│ Verify Slack #tech │
│ Verify Sentry token │
└───────┬──────────────────┘
▼
┌─ STAGE -1: Reconcile ────┐ → reference/reconcile.md
│ List merged sentry- │ reference/sentry-api.md
│ labeled PRs since last │
│ run; resolve still- │
│ unresolved Sentry IDs │
└───────┬──────────────────┘
▼
┌─ STAGE 0: Changelog scan ┐ → reference/reconcile.md (same file)
│ Pre-resolve fixes that │
│ already shipped │
└───────┬──────────────────┘
▼
┌─ STAGE A: Triage ────────┐ → reference/triage.md
│ Fetch ALL unresolved │
│ Explore agents in │
│ parallel (one per │
│ cluster, top 10 only) │
└───────┬──────────────────┘
▼
┌─ RANK & GATE ────────────┐ → reference/ranking.md
│ Top 3 by confidence, │
│ scope, freq. Gate by │
│ confidence × scope. │
│ Gated → GH issues. │
└───────┬──────────────────┘
▼
┌─ STAGE B: Fix (K=3) ─────┐ → reference/dispatch.md
│ One zellij pane per │ reference/raffle.md
│ auto-dispatchable issue │
│ running /spec │
│ --auto-approve │
└───────┬──────────────────┘
▼
┌─ CYCLE ANNOUNCEMENTS ────┐ → reference/slack.md
│ Msg 1 (cycle started) │
│ Msg 2 (PRs ready + │
│ raffle) after CI │
│ Msg 3 (resolved on │
│ merge, via Stage -1 │
│ of NEXT run) │
└──────────────────────────┘
Read only what you need for the current stage:
| Stage / topic | Read |
|---|---|
| Env vars, Slack MCP load, multiplexer check | reference/prereqs.md |
| Sentry API URL forms + canonical resolve call | reference/sentry-api.md |
| Stage -1 reconcile + Stage 0 changelog scan | reference/reconcile.md |
| Stage A triage (cluster, fan-out, Explore briefs) | reference/triage.md |
| Rank/gate rules + GH dashboard writeup | reference/ranking.md |
| Stage B dispatch (plan file, zellij, PR label) | reference/dispatch.md |
| Reviewer raffle, self-nomination, author-collapse | reference/raffle.md |
#tech message templates (1/2/3) + channel filter | reference/slack.md |
Single-issue mode flow + <ref> resolution | reference/single-mode.md |
/sentry-fix [project1,project2]: reconcile, changelog scan, triage, top 3, dispatch./sentry-fix dry-run: runs Stage A only, posts what would dispatch, skips Stage B./sentry-fix single <ref>: targeted dispatch, skips triage-the-world. See reference/single-mode.md.This skill does NOT auto-queue. After dispatching the top 3 and posting Message 1, it stops. To process the next 3 unresolved issues, the user re-runs /sentry-fix. Natural cadence: 3 PRs land, user triages them, next run picks the next 3.
dev merge, not on prod. If a fix doesn't actually work in prod, Sentry will re-open the issue naturally.reference/triage.md), but two Sentry issues in different cycles each get their own spec loop. Fine for now; revisit if noisy./diagnose — when a Sentry issue is gated as low-confidence and you want to attach manually rather than let /spec auto-dispatch, use /diagnose for the disciplined repro-and-root-cause loop before fixing.