Przejdź do treści
Biblioteka skilli
DevOps i chmura · Integracja z narzędziem

gluesync-core-hub

/gluesync-core-hub

Obsługuje pipeline'y replikacji Gluesync Core Hub: źródła, cele, mapowanie encji, migawki i CDC oraz diagnozuje zatrzymania.

Licencja
Apache-2.0
Ostatnia zmiana repozytorium
11.11.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
8
Skrypty
3
Komentarze HTML
0
Adresy URL w SKILL.md
14
Przestarzałość

Brak sygnałów. Repozytorium zmienione 11.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

Co sprawdzić przed instalacją
  • 3 skrypty w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.

Skrypty: scripts/gs_api.py, scripts/gs_common.py, scripts/mcp.py

Co robi i kiedy

Obsługuje pipeline'y replikacji Gluesync Core Hub: źródła, cele, mapowanie encji, migawki i CDC oraz diagnozuje zatrzymania.

Oryginalny opis (EN)

Work with a Gluesync Core Hub (MOLO17's real-time data replication and CDC platform): create and run pipelines, add source and target agents, map entities, start snapshots and CDC, check status and metrics, and diagnose stalled or failing replication, through the Core Hub REST API, its built-in MCP server, or step-by-step guidance in the Control Plane web UI. Use whenever the user mentions Gluesync, Core Hub, Control Plane, a Gluesync pipeline, entity or agent, a gsp_ token, or asks to replicate or sync data between databases with Gluesync.

Jak działa

  1. Pick the route
  2. Before calling anything
  3. Rules that keep replication data safe
  4. Build a pipeline end to end
  5. Answering "is it working?"
  6. Where to send the user
Sekcje z nagłówków SKILL.md, w oryginale · 110 linii, 921 słów
Podgląd SKILL.md
---
name: gluesync-core-hub
description: >-
  Work with a Gluesync Core Hub (MOLO17's real-time data replication and CDC platform): create and
  run pipelines, add source and target agents, map entities, start snapshots and CDC, check status
  and metrics, and diagnose stalled or failing replication, through the Core Hub REST API, its
  built-in MCP server, or step-by-step guidance in the Control Plane web UI. Use whenever the user
  mentions Gluesync, Core Hub, Control Plane, a Gluesync pipeline, entity or agent, a gsp_ token,
  or asks to replicate or sync data between databases with Gluesync.
---

# Gluesync Core Hub

Gluesync replicates data between databases in real time: a **pipeline** connects a **source agent**
(reads, usually by change data capture) to one or more **target agents** (writes), and each table
or collection it carries is an **entity**. Core Hub is the server that runs it all; the **Control
Plane** is its web UI. Everything here is grounded in the official documentation at
<https://docs.molo17.com/gluesync/latest/>. When this skill and the user's Core Hub disagree, the
Core Hub wins: releases differ, so check its live Swagger (`<core-hub>/swagger`) or the MCP tool
list rather than insisting.

## Pick the route

| The user wants to… | Use |
|---|---|
| Be guided through a task, or has no API access | The Control Plane UI: [references/ui.md](references/ui.md) |
| Let Claude act on their Core Hub, and it is connected as an MCP server | The MCP tools directly: [references/mcp.md](references/mcp.md) |
| Let Claude act, with no MCP connector | `scripts/mcp.py` (MCP over the command line) or `scripts/gs_api.py` (REST) |
| Script or automate outside Claude | REST: [references/rest-api.md](references/rest-api.md) |
| Fix something that is stuck or failing | [references/troubleshooting.md](references/troubleshooting.md) |

Prefer MCP over raw REST when both are available: its tools encode the order of operations and
the payload shapes, and `validate_entities` checks an entity before anything is written.

## Before calling anything

1. **Ask for the Core Hub address.** It is the HTTPS address from installation: `https://<host>`
   behind the bundled reverse proxy (port 443), or `https://<host>:1717` without it. TLS is on by
   default, often with a self-signed certificate on trial kits.
   ([TLS](https://docs.molo17.com/gluesync/latest/security/setup-https-and-wss-encryption.html),
   [REST](https://docs.molo17.com/gluesync/latest/core-hub/core-hub-apis.html))
2. **Get a Personal API Token, with the narrowest scope that does the job.** The user creates it in
   the Control Plane: user menu → **Settings** → **Account** → **Personal API Tokens** →
   **New token**; it starts with `gsp_` and is shown once. Scopes: `gluesync.mcp_read` or
   `gluesync.api_read` to look, `gluesync.mcp` or `gluesync.api` to change things. MCP scopes do not
   open REST, and API scopes do not open MCP.
   ([Personal API Tokens](https://docs.molo17.com/gluesync/latest/security/personal-api-tokens.html))
3. **Never ask the user to paste a token or password into the chat.** Have them export it in their
   own shell (`export GLUESYNC_PAT=...`) so the scripts read it from the environment. Never echo,
   log or write secrets to a file.
4. Set the environment for the scripts:
   ```bash
   export GLUESYNC_BASE=https://gluesync.example.com      # the user's Core Hub
   export GLUESYNC_PAT=gsp_...                             # set by the user, not by you
   # self-signed certificate: export GLUESYNC_CA_FILE=/path/ca.pem   (or GLUESYNC_INSECURE_TLS=1 on a test box)
   python3 scripts/gs_api.py whoami                        # REST scope check
   python3 scripts/mcp.py tools                            # MCP scope check, lists the tools
   ```

## Rules that keep replication data safe

- **Read first, then change.** Show the user what you will create, start, stop or delete, and get
  a yes before any write: pipelines, agents, connections, entities, sync commands, maintenance.
- **Never without the user's explicit, informed yes:** deleting pipelines, entities or connections;
  **Empty the table first** / **Delete only some rows first**; `redo` / restart from zero.
- **Reset pipeline checkpoint and Restart platform are support-supervised operations.** Do not run
  them; tell the user to involve Gluesync support
  ([stalled pipelines](https://docs.molo17.com/gluesync/latest/learn/how-to-diagnose-a-stalled-pipeline-or-entity.html)).
- **A 202 means dispatched, not done.** After a command, read the state back
  (`get_pipeline_status` / `GET /pipelines/{id}/entities-status`, metrics) before saying it worked.
  Retries are not idempotent; check before re-sending.
  ([REST](https://docs.molo17.com/gluesync/latest/core-hub/core-hub-apis.html))
- **Commands take entity IDs, not names.** A name is accepted with 202 and does nothing. (The MCP
  `sync_command` tool, where present, accepts `entity_names`.)
  ([automate how-to](https://docs.molo17.com/gluesync/latest/learn/how-to-automate-gluesync-with-the-rest-api-mcp-and-webhooks.html))
- **Green is not proof.** An entity can sit Active with "Idle, no changes yet" while its database
  is down. Prove delivery with a probe row on the source, then look for it on the target.
- **Error text can contain row values.** Treat notifications, logs and copied errors as personal
  data: quote only what is needed.

## Build a pipeline end to end

Through MCP, the order is fixed; each step depends on the previous one:

1. `create_pipeline` → 2. `add_agent` (source, then target; `agent_internal_name` is the agent tag,
see `list_agent_types`) → 3. `assign_agent` (without it, later steps fail) → 4. `create_connection`
→ 5. `attach_connection` (pushes credentials; must precede discovery) → 6. leave Draft with
`complete_pipeline_configuration` where the tool exists → 7. `set_agent_specific_config` if needed
→ 8. `draft_entity` (runs discovery and maps columns) → 9. `validate_entities` → 10.
`upsert_entities` → 11. start: `sync_command {"command": "start", "with_snapshot": true}` on newer
releases, `start_sync` on earlier ones.

`scripts/mcp.py setup` performs exactly that sequence and prints each step; confirm the plan with
the user first, and pass database passwords via `GLUESYNC_SOURCE_PASSWORD` /
`GLUESYNC_TARGET_PASSWORD`. Run `python3 scripts/mcp.py schema <tool>` before calling a tool whose
arguments you have not seen on this Core Hub.

Then follow a change through: insert, update **and** delete a probe row on the source and confirm
each on the target. Deletes and updates are where mapping problems show up.

In the UI the same job is the **Pipeline** wizard (Setup source → Setup target(s) → Ready), then
**Entities** → Objects Browser → Fields Editor → **Start sync**. See [references/ui.md](references/ui.md).

## Answering "is it working?"

1. `get_pipeline_status` (MCP) or `GET /pipelines/{id}/entities-status` (REST): look at
   `errorState`, `isSyncActive`, `isMigrationActive`.
2. `get_system_conditions` where available: problems affecting the whole Core Hub (`[]` = healthy).
3. Notifications: `get_notifications`, or the bell icon → **Notifications hub**.
4. Snapshot progress: `get_pipeline_snapshot_status`; for counts use **Count both** or a SQL count,
   because the percentage is approximate.
5. Then the probe row. Diagnose with [references/troubleshooting.md](references/troubleshooting.md).

## Where to send the user

- Documentation: <https://docs.molo17.com/gluesync/latest/>
- Support tickets: <https://support.molo17.com>
  ([how to open one](https://docs.molo17.com/gluesync/latest/learn/how-to-create-a-support-ticket.html))
- Service status: <https://status.molo17.com/>
Pliki w paczce (8)
  • SKILL.md 8 KB
  • references/mcp.md 6 KB
  • references/rest-api.md 6 KB
  • references/troubleshooting.md 5 KB
  • references/ui.md 7 KB
  • scripts/gs_api.py 3 KB
  • scripts/gs_common.py 4 KB
  • scripts/mcp.py 10 KB

Wersje

W bazie nie ma innych skilli o nazwie gluesync-core-hub.

Migawka bazy z 13.11.2026 · wersja paczki b1117c066c34 (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100cdc-pipeline-creator
    Prawdopodobnie tworzy potoki CDC (change data capture) w ramach pracy z potokami danych.
  • 100nf-core-pipeline-create
    Tworzy szkielet nowego pipeline'u z szablonu nf-core poleceniem nf-core pipelines create, z domyślnymi metadanymi i pierwszą weryfikacją.
  • 90hub-init
    Tworzy nowy hub de-agent dla zespołu: wybiera repozytoria i źródła danych, przygotowuje konfigurację bez sekretów i mapę.
  • 100b2c-replication
    Automatyzuje replikację kodu i danych z instancji staging B2C Commerce na produkcję lub dev przez zadanie CLI b2c.
  • 100operating-guide
    Przedstawia zasady pracy z Control Plane: fakty o platformie, zatwierdzenia o dużym wpływie, sekrety, cele i awarie.
  • 100nf-core-pipeline-schema
    Dodaje, zmienia lub usuwa parametr pipeline'u i regeneruje nextflow_schema.json poleceniami nf-core pipelines schema.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.