add-source
/add-source
Dodaje do huba system na żywo, którego jeszcze nie monitoruje, np. Snowflake, BigQuery lub MySQL, z blokiem źródła i sekretami tylko przez referencje.
- Autor
- Rafid-Bin-Sadeque ↗
- Repozytorium
- de-agent ↗
- Licencja
- Apache-2.0
- Ostatnia zmiana repozytorium
- 03.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 1
Brak sygnałów. Repozytorium zmienione 03.11.2026, bez odwołań do wycofanych modeli i API.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
Co robi i kiedy
Dodaje do huba system na żywo, którego jeszcze nie monitoruje, np. Snowflake, BigQuery lub MySQL, z blokiem źródła i sekretami tylko przez referencje.
Wyzwalacze z opisu
Oryginalny opis (EN)
Add a live system the hub does not monitor yet (Snowflake, BigQuery, Databricks, MySQL, SQL Server, GCP, Kubernetes, Dagster, Fivetran, or anything with SQL, a CLI or an HTTP API). Finds an existing pack or drafts one, checks it, proposes the de-agent.yaml source block with secret references only, tests it read-only and iterates. Use when the user says "we also use X" or "monitor X".
Jak działa
- When to use
- Inputs
- Steps
- Output
- Never
Podgląd SKILL.md
---
name: add-source
description: Add a live system the hub does not monitor yet (Snowflake, BigQuery, Databricks, MySQL, SQL Server, GCP, Kubernetes, Dagster, Fivetran, or anything with SQL, a CLI or an HTTP API). Finds an existing pack or drafts one, checks it, proposes the de-agent.yaml source block with secret references only, tests it read-only and iterates. Use when the user says "we also use X" or "monitor X".
---
# add-source
## When to use
The user names a system that is not in `de-agent.yaml` `sources:` ("we also run a MySQL database for reviews", "we use Snowflake").
The system has SQL, a command-line client or an HTTP API. A *pack* (a small YAML file) describes what to read; no code is written.
## Inputs
- Which system, and which parts of it to watch (named resources only: warehouses, datasets, namespaces, job ids, groups).
- How the connection is made (host or URL, role or user) and HOW the secret is provided: `env:VAR`, `file:/abs/path` (mode 0600) or
`none`. Never ask for the secret value; ask the human to put it in their environment or `.env` themselves.
## Steps
1. `dea pack list`. If a pack fits (`mysql`, `snowflake`, `bigquery`, `databricks`, `databricks-system`, `mssql`, `gcp`, `kubernetes`,
`dagster`, `fivetran`, `postgres-lite`), use it: `dea pack show <name>` to see its params and sections. Note its `validated:` level
(`fixtures` means it was never run against a live system: say so).
2. No pack fits: `dea pack new <name> --kind sql|cli|http`, then draft the sections in `packs/<name>.yaml` from the system's DOCUMENTED
system views, CLI or API (use the explorer agent or the docs you already know; do not invent tables or flags). Every SQL is one
`select`/`show`; every `run:` is a read verb of a known CLI; every http section is a `get:` or a GraphQL `query`. Scope each section
to named resources through `params:` (a list the user supplies); no account-wide listings.
3. `dea pack check <name>`. Fix every reported problem; a pack that fails the check never runs.
4. Propose the `de-agent.yaml` source block (show the diff, then add it). Secret fields hold references only:
```yaml
- name: reviews-db
type: sql # or cli / http
pack: mysql
conn: {url: "mysql+pymysql://monitor@reviews-db.internal:3306/reviews", password: "env:REVIEWS_DB_PASSWORD"}
options: {} # the pack's params
```
Ask the human to export the variable (or put it in the hub's `.env`) and install the extra the pack needs
(`pip install 'de-agent[mysql]'`; `dea doctor` names it).
5. `dea doctor`, then `dea pack test <source>`: it runs the source once, read-only, and prints the rendered page and the signals.
6. Iterate: fix a section that errors, tighten a threshold, add a signal. Re-run `dea pack check` and `dea pack test`. When it is
good, `dea snapshot` and `dea lint`.
## Output
- The pack used or created (name, kind, `validated:` level), the source block added, `dea pack check` and `dea pack test` results,
and anything the human still has to do (export a secret, install an extra, grant a read-only role: point to the matching block in
`docs/read-only-access.md` in the de-agent plugin folder, online at
https://github.com/Rafid-Bin-Sadeque/de-agent/blob/main/docs/read-only-access.md).
## Never
- Never put a write statement or a mutating verb in a pack (`insert`, `delete`, `create`, `kubectl delete`, `run-now`, a POST mutation).
The runtime refuses them, but do not write them.
- Never write a password, token or key into any file or the transcript: `env:` / `file:` references only.
- Never list a whole account, cluster or project when named resources will do.
- Never mark `validated: live` unless the pack was really run against that system; fixtures-only packs say `fixtures`.
- Never run `dea approve` or `dea apply`; adding a source is read-only.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · Rafid-Bin-Sadeque/de-agent | 100 | Apache-2.0 |
| mmca950/assara | 100 | MIT |
| leclaudeur/hodi | 100 | AGPL-3.0 |
| tmargolis/career-navigator | 100 | Apache-2.0 |
| dividor/athetize-plugin-claude | 90 | MIT |
| howdycarter/role-mosaic | 90 | MIT |
| ivanwng97/pixtuoid | 90 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 8ed2e1a284df (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
- 100sql
Sprawdza, poprawia i przepisuje zapytania SQL w PostgreSQL, Spark SQL, T-SQL, BigQuery i Snowflake, także pod kątem kosztu. - 100generating-sql
Zamienia pytania biznesowe na poprawny SQL dla PostgreSQL, BigQuery, Snowflake i innych baz, wyjaśnia zapytanie i poprawia istniejące zapytania. - 100malloy
Pomaga pisać modele, źródła, widoki, złączenia i zagnieżdżone zapytania w Malloy, języku modelowania semantycznego kompilowanym do SQL dla DuckDB, BigQuery, Snowflake, PostgreSQL i innych baz. - 100snowflake-to-bigquery-converter
Przekształca SQL w projektach LookML z dialektu Snowflake na BigQuery Standard SQL przez looker-cli. - 100cloud-sql-basics
Tworzy lub wyjaśnia instancje Cloud SQL dla MySQL, PostgreSQL i SQL Server, których kopie zapasowe i dostępność zarządza Google. - 100sql-expert
Pomaga pisać i optymalizować zapytania SQL oraz projektować bazy w PostgreSQL, MySQL, SQLite i BigQuery.