Connect Claude, ChatGPT and any AI agent to jentry

jentry is MCP-first. Paste one URL into your AI app, sign in, click Allow — and ask “what broke since my deploy?”, “show me the stack trace”, “mark it resolved” or “set up error tracking for this new app”. The agent acts as you, with your role, and asks before anything it can't undo.

MCP server URL

https://jentry.app/mcp

97 tools · 44 read-only · 53 that change things ask first · OAuth 2.1, nothing to install

Set up

Connect your client

Every client below signs in with OAuth: it opens jentry, you log in and click Allow. Nothing to register anywhere, and you can disconnect it any time in Settings → AI agents.

Claude

claude.ai · Claude Desktop · Claude mobile

1In claude.ai or Claude Desktop, open Settings → Connectors.
2Click Add custom connector, name it jentry and paste the server URL.
3Click Connect. jentry opens: sign in and click Allow.
4In a chat, turn jentry on from the tools menu and ask. Connectors you add on claude.ai also show up in the Claude apps for iOS and Android.

On Team and Enterprise plans an owner adds the connector once in the organization's settings; then each member clicks Connect and signs in with their own jentry account.

Claude Code

Terminal and IDE extensions

Add the server

claude mcp add --transport http jentry https://jentry.app/mcp

1Run the command above (add --scope user to have jentry in every project).
2Inside Claude Code, type /mcp, pick jentry and choose Authenticate.
3Your browser opens jentry: sign in and click Allow. Back in the terminal, ask away.

ChatGPT

Apps & Connectors, developer mode

1Open Settings → Apps & Connectors → Advanced settings and turn on Developer mode.
2Back in Apps & Connectors, click Create.
3Name it jentry, paste the URL as the MCP server URL and choose OAuth for authentication. Confirm you trust it and click Create.
4jentry opens: sign in and click Allow.
5In a new chat, add jentry from the + menu and ask.

On ChatGPT Business and Enterprise workspaces an admin may have to allow developer mode or custom connectors first.

Cursor

mcp.json

1Open Cursor Settings → MCP (Tools & MCP) and click New MCP Server. That opens ~/.cursor/mcp.json (or use .cursor/mcp.json in a project).
2Add jentry and save:

~/.cursor/mcp.json

{
  "mcpServers": {
    "jentry": { "url": "https://jentry.app/mcp" }
  }
}

jentry then shows Needs login: click it, sign in to jentry and click Allow. The tools appear in Cursor's agent.

VS Code

GitHub Copilot agent mode

1Run MCP: Add Server… from the Command Palette, choose HTTP, paste the URL and name it jentry. Or add it to .vscode/mcp.json:

.vscode/mcp.json

{
  "servers": {
    "jentry": { "type": "http", "url": "https://jentry.app/mcp" }
  }
}

Start the server; VS Code asks to sign in to jentry — allow it, log in and click Allow. In Copilot Chat, switch to Agent mode and the jentry tools are there.

Any MCP client

Streamable HTTP + OAuth 2.1

Use the URL as a Streamable HTTP server. Clients that speak MCP authorization find everything on their own: the 401 from /mcp points at /.well-known/oauth-protected-resource/mcp, jentry accepts dynamic client registration, and PKCE (S256) is required. Redirects may go to https, a loopback port on your computer, or a known app's own callback (Cursor's). Clients that can only send a header use a manual token (below).

No OAuth?

Manual tokens

For scripts, CI agents or a client without OAuth: create a token in Settings → AI agents (shown once, up to 20) and send it as Authorization: Bearer jmcp_…. It acts as you, like an OAuth connection, and is revoked from the same page.

Claude Code with a token

claude mcp add --transport http jentry https://jentry.app/mcp --header "Authorization: Bearer jmcp_…"

mcp.json with a token (Cursor and others)

{
  "mcpServers": {
    "jentry": {
      "url": "https://jentry.app/mcp",
      "headers": { "Authorization": "Bearer jmcp_…" }
    }
  }
}

Agent tokens are per person and work in all your organizations. They are not the organization's REST API tokens nor a project's DSN, and jentry stores only their hash.

Capabilities

What an agent can do

Everything you can do in the dashboard, with your role in each organization — 97 tools in 11 groups. Read-only tools run freely; tools that change something are announced as such, so Claude and other clients ask you before each call.

reads changes something (your client asks first)

Start here & projects

5 tools

What needs attention right now, your organizations and projects, a DSN plus install code for a new app, and importing from Sentry.

get_overview (read-only)
list_projects (read-only)
get_project_setup (read-only)
create_project (changes something)
import_from_sentry (changes something)

Releases & deploys

3 tools

Releases with their commits, deploys and new issues; register a release from a chat or a CI agent.

list_releases (read-only)
get_release (read-only)
create_release (changes something)

Issues & triage

22 tools

Search issues, read stack traces and breadcrumbs, find what broke since a deploy, then resolve, ignore, assign, merge or file them in GitHub/GitLab.

list_issues (read-only)
get_issue (read-only)
list_issue_events (read-only)
issues_since (read-only)
get_issue_tags (read-only)
list_similar_issues (read-only)
get_error_stats (read-only)
resolve_issue (changes something)
ignore_issue (changes something)
reopen_issue (changes something)
assign_issue (changes something)
comment_issue (changes something)
update_issues (changes something)
set_issue_priority (changes something)
merge_issues (changes something)
unmerge_issue (changes something)
delete_issues (changes something)
analyze_issue (changes something)
create_external_issue (changes something)
list_saved_searches (read-only)
create_saved_search (changes something)
delete_saved_search (changes something)

Alerts & notifications

7 tools

Alert rules for new issues, spikes and metrics, pausing them, and your in-app notifications.

list_alert_rules (read-only)
create_alert_rule (changes something)
update_alert_rule (changes something)
set_alert_rule_enabled (changes something)
delete_alert_rule (changes something)
list_notifications (read-only)
mark_notifications_read (changes something)

Your account

4 tools

Your account, alert e-mail preferences and the AI apps connected to it.

get_account (read-only)
set_notification_preferences (changes something)
list_agent_connections (read-only)
revoke_agent_connection (changes something)

Performance & tracing

15 tools

Transactions, p95 and Apdex, traces, the service map, DB/HTTP/cache insights, profiles, replays, custom metrics, user feedback, Discover and coverage.

get_performance (read-only)
get_transaction (read-only)
search_traces (read-only)
get_trace (read-only)
get_service_map (read-only)
get_insights (read-only)
list_profiles (read-only)
get_profile (read-only)
list_replays (read-only)
get_replay (read-only)
get_metrics (read-only)
list_feedback (read-only)
run_discover (read-only)
list_coverage_reports (read-only)
get_coverage_report (read-only)

Crons, uptime & dashboards

12 tools

Cron monitors and their check-ins, uptime checks, and custom dashboards with their widgets.

list_cron_monitors (read-only)
create_cron_monitor (changes something)
delete_cron_monitor (changes something)
list_uptime_checks (read-only)
create_uptime_check (changes something)
delete_uptime_check (changes something)
list_dashboards (read-only)
get_dashboard (read-only)
create_dashboard (changes something)
add_dashboard_widget (changes something)
delete_dashboard (changes something)
delete_dashboard_widget (changes something)

Project settings

10 tools

DSN keys and rate limits, inbound filters and data scrubbing, GitHub/GitLab links, ownership rules and code mappings.

get_project_settings (read-only)
update_project_filters (changes something)
create_client_key (changes something)
update_client_key (changes something)
delete_client_key (changes something)
set_project_integration (changes something)
create_ownership_rule (changes something)
delete_ownership_rule (changes something)
create_code_mapping (changes something)
delete_code_mapping (changes something)

Organization & team

13 tools

Organizations, members and invitations, roles, API tokens (list and revoke), the audit log and the Seer AI key.

get_organization (read-only)
create_organization (changes something)
accept_invitation (changes something)
list_members (read-only)
list_invitations (read-only)
invite_member (changes something)
revoke_invitation (changes something)
change_member_role (changes something)
remove_member (changes something)
list_api_tokens (read-only)
revoke_api_token (changes something)
list_audit_log (read-only)
set_seer_key (changes something)

Plan & usage (links only)

3 tools

Usage against your plan. Upgrades come back as a Stripe Checkout link you pay yourself, and the billing portal as a link: an agent never pays or charges anything.

get_billing (read-only)
change_plan (changes something)
open_billing_portal (changes something)

Everything else: the REST API

3 tools

jentry’s REST API (/api/v1) as you, with your role’s permissions. Reads and writes are separate tools, and writes show a preview first.

api_get (read-only)
api_create (changes something)
api_update (changes something)

Clients that support MCP prompts also get 3 ready-made workflows: triage_issues, weekly_health_review and setup_new_project.

Examples

Try asking

“What needs attention in jentry right now?”

Services down, failing crons, regressions and new or spiking issues, each with the next step.

“What broke since we deployed web@2.4.0? Show me the stack trace of the worst one.”

New and regressed issues for that release, with the in-app frames.

“Set up error tracking for my new Next.js app.”

Creates the project and returns the DSN plus the install and init code.

“Resolve every issue in checkout that hasn’t happened in 14 days.”

Bulk triage, recorded under your name.

“E-mail me when a new issue shows up in production for the api project.”

An alert rule; e-mail and webhook targets are shown before they’re saved.

“Give me a weekly health review of our organization.”

Errors, slow endpoints, crons, uptime and quota, with what to do next.

Trust

Safety & privacy

It acts as you — nothing more

Only the organizations you belong to, with your role in each. Leave an organization and the agent loses it too. Every change is recorded under your name, marked as made through an agent.

Reads run, changes ask

Read-only tools are labeled read-only. Every tool that changes something is labeled as such, so Claude, ChatGPT and other clients ask you before each call.

Previews before anything big

Deleting, merging, revoking, removing members and anything that leaves jentry (invitations, webhooks, GitHub/GitLab issues, Sentry imports) first returns a preview of what would happen and only runs once you confirm.

No money moves

No tool pays, charges or refunds anything. A plan upgrade comes back as a Stripe Checkout link you open and pay yourself; invoices, cards and cancellations stay in the billing portal.

Credentials stay in the browser

An agent can’t create or reveal tokens, DSN secret keys, your password or 2FA, and answers never include secrets.

Revoke in one click

Settings → AI agents lists every connected app and token: Disconnect or Revoke ends it at once, and a password reset ends them all. OAuth access tokens last an hour; refresh tokens rotate on every use.

What data flows. What an agent reads goes to the AI app you connected, under that app's terms. jentry trims it first: e-mails are masked, and IPs, cookies, auth headers, request bodies and stack-frame variables are left out. jentry never receives your conversation — only the tool calls the agent makes. Error messages and breadcrumbs are written by whoever sends events to your DSN, so agents are told to treat them as data, never as instructions. Details in the privacy policy.

FAQ

Frequently asked questions

Does jentry have an MCP server?

Yes. jentry is MCP-first: point any MCP client at https://jentry.app/mcp. It signs in with OAuth (you log in to jentry and click Allow) or with a token from Settings → AI agents.

Do I need to install anything?

No. It is a remote server (Streamable HTTP). Claude, ChatGPT, Cursor and VS Code only need the URL; Claude Code needs one command.

Can the agent delete things or spend money on its own?

No. Deleting, merging, revoking and anything that leaves jentry first returns a preview and only runs after you confirm. No tool pays, charges or refunds anything: a plan upgrade comes back as a Stripe Checkout link you open and pay yourself.

What does the agent see?

What you can see, in the organizations you belong to, with your role. Event data has e-mails masked and IPs, cookies, auth headers and request bodies removed before it reaches the agent.

How do I disconnect an agent?

Settings → AI agents lists every connected app and token; Disconnect or Revoke cuts it off at once. Resetting your password revokes them all.

Stuck connecting?

Write to sales@devcloudsoftware.com with the client you use and what you see. Don't have an account yet? It's free to start.

Get started — free Settings → AI agents