// Use the official Sentry CLI to investigate issues, events, traces, and logs for Vibe Remote without committing Sentry credentials into the repository.
Use `vibe watch` to run a managed Harness waiter that returns to the same conversation later. Best for reviews, CI, files, logs, and other wait-now-continue-later workflows.
Safely inspect and modify local Vibe Remote configuration, routing, runtime settings, watches, scheduled tasks, Vibe Cloud remote access, and operational state.
| name | use-sentry-cli-debug |
| slug | use-sentry-cli-debug |
| description | Use the official Sentry CLI to investigate issues, events, traces, and logs for Vibe Remote without committing Sentry credentials into the repository. |
| version | 0.1.0 |
Use this skill when the user asks to inspect Sentry issues, events, traces, logs, or authenticated Sentry API responses from the command line while working in the Vibe Remote repository.
sentry CLI subcommands over raw REST calls. Use sentry issue, sentry event, sentry trace, and sentry log first. Fall back to sentry api only when the dedicated command cannot expose the needed field.sentry auth login interactive auth. If token auth is required, use a shell-scoped SENTRY_AUTH_TOKEN or another untracked local secret source. Never write tokens into tracked repo files.--json --fields ..., --limit ..., and jq when filtering.sentry auth status, then confirm the target org/project if auto-detection looks ambiguous.If sentry is not installed:
curl https://cli.sentry.dev/install -fsS | bash
Then verify:
sentry --help
sentry auth status
Authentication options that do not require committing secrets:
# Preferred: interactive login managed by the CLI
sentry auth login
# Alternative: shell-scoped token
export SENTRY_AUTH_TOKEN=...
sentry auth login --token "$SENTRY_AUTH_TOKEN"
Never add the token to tracked files such as .env, .envrc, config/*.json, settings.json, or skill files.
Before querying Sentry, read vibe-remote.md for the tags and environment metadata currently attached by this repository.
Important current signals:
component=service for the main servicecomponent=ui for the Flask UImode=<config.mode>primary_platform=<config.platforms.primary>deployment_environment=<resolved environment>release=vibe-remote@<version>These are useful narrowing dimensions when many unrelated issues exist in the same Sentry org/project.
sentry auth status
sentry org list
sentry project list <org>
If auto-detection is unreliable, use explicit <org>/<project> in every command.
sentry issue list <org>/<project> \
--query 'is:unresolved' \
--limit 10 \
--json \
--fields shortId,title,count,userCount,lastSeen,level,status
Common filters:
sentry issue list <org>/<project> --query 'is:unresolved environment:regression'
sentry issue list <org>/<project> --query 'component:service'
sentry issue list <org>/<project> --query 'component:ui'
sentry issue list <org>/<project> --query 'primary_platform:slack'
sentry issue list <org>/<project> --query 'release:vibe-remote@1.2.3'
sentry issue view <short-id>
sentry issue explain <short-id>
Use sentry issue explain only after you have the raw issue details. Treat the AI explanation as a hint, not ground truth.
sentry event view <org>/<project>/<event-id>
sentry trace view <trace-id>
sentry trace logs <trace-id> --limit 100
sentry log list <org>/<project> --query 'level:error' --limit 100
Use trace and log commands when the issue points to timing, retries, upstream HTTP failures, or multi-step orchestration problems.
sentry api only as a fallbacksentry api /api/0/organizations/<org>/issues/
Prefer --method, --data, and --header only when the dedicated CLI command cannot express the operation.
environment:regression.component:service and component:ui early. Do not mix backend startup errors with UI Flask failures.primary_platform:<platform> and cross-check local logs if Sentry detail is incomplete.~/.vibe_remote/logs/vibe_remote.log.sentry CLI commands directly without a full debug workflow.