A silo works best with one writer (you, a script, or a CI job) and many readers. The Manage API isn’t built for many users writing to the same silo at the same time. See Update Silo.
Setup
.env
Create a silo
file_data and region_name are required. The response contains the new silo’s file_uuid.
Response
Change one value
file_data in an update replaces the whole document. To change one value, read the silo, change it in your code, and send all of it back.
This script sets the stock of one product:
When the schema says no
If the silo has a locked schema, an update that doesn’t match it returns422 and nothing is saved. For example, sending "stock": "35" (a string) to a silo whose schema expects a number returns:
422 Response
path is a JSON Pointer to the first invalid value. Fix it and send the request again.
Sync a file from GitHub Actions
Keep the JSON in your repository, review changes in pull requests, and push it to the silo when it lands onmain.
1
Add the file to your repo
For example,
config/flags.json.2
Add a secret and a variable
In your repository, open Settings → Secrets and variables → Actions and add:
- A secret named
JSONSILO_MANAGE_KEYwith your Manage Key. - A variable named
JSONSILO_SILO_UUIDwith the silo’s UUID.
3
Add the workflow
.github/workflows/sync-jsonsilo.yml
api in the URL if your silo is in another region.422 when the file doesn’t match the silo’s locked schema. That makes Schema Lock a useful safety net for this workflow: a broken file never reaches the apps reading the silo.
Good to know
- Changes are logged. Every create, update, and delete through the Manage API appears in Logs with an (API) label.
- Watch the headers. Responses include
ManageSilo-Req-Remaining, so a script can slow down before it hits the rate limit. - Changing
region_nameoris_publicchanges the silo’s URL. Only send them when you mean to.