Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/org-quicko/silo/llms.txt

Use this file to discover all available pages before exploring further.

Silo’s filter is a JSON AST, not a string language. You describe exactly what you want, and Silo validates the shape before touching storage. Fields are addressed using RFC 9535 JSONPath over a document of {id, rev, created_at, updated_at, data} — your own fields always live under $.data, so a field named id in your schema can never shadow the envelope’s $.id.

​
JSONPath addressing

Every filter, sort, and search path is a JSONPath expression over the entry document.
PathSelects
$.idEntry ULID
$.revCurrent revision number
$.created_atCreation timestamp (RFC 3339 UTC)
$.updated_atLast-updated timestamp (RFC 3339 UTC)
$.data.titleA top-level field in your schema
$.data.author.nameA nested field
$.data.tags[*]All elements of an array
$.data.tags[0]First element of an array (negative indices work too)
Supported: root ($), name selectors (.field), array indices ([n], negative included), and the child wildcard ([*]). Not supported (actively refused): recursive descent (..), slices ([0:2]), unions ([a,b]), filter selectors ([?…]), and function extensions. Silo refuses these by name rather than silently ignoring them, so a typo or unsupported construct is an error, not a silent no-op.

​
Filter AST

A filter is a JSON object with an op field. Logical operators nest leaf conditions. 
{ "op": "eq",       "path": "$.data.status",  "value": "published" }
{ "op": "neq",      "path": "$.data.status",  "value": "draft" }
{ "op": "gt",       "path": "$.data.score",   "value": 5 }
{ "op": "gte",      "path": "$.data.score",   "value": 5 }
{ "op": "lt",       "path": "$.data.score",   "value": 10 }
{ "op": "lte",      "path": "$.data.score",   "value": 10 }
{ "op": "in",       "path": "$.data.status",  "value": ["published", "archived"] }
{ "op": "contains", "path": "$.data.title",   "value": "silo" }
{ "op": "exists",   "path": "$.data.summary" }
A leaf is true when any node the path selects satisfies it. Any over nothing is false.
Array semantics are easy to get wrong. neq($.data.tags[*], "x") means some tag is not “x” (true if there is at least one tag that differs). not(eq($.data.tags[*], "x")) means no tag is “x” (true only if the value is absent from every element). Use not(eq(…)) when you want strict exclusion.

​
Compound example

{
  "op": "and",
  "args": [
    { "op": "eq",       "path": "$.data.status",      "value": "published" },
    { "op": "contains", "path": "$.data.author.name", "value": "ada" },
    { "op": "eq",       "path": "$.data.tags[*]",     "value": "release" }
  ]
}
This matches entries that are published, written by someone whose name contains “ada”, and tagged with “release”.

​
Using a filter in a request

Pass the filter JSON as a URL-encoded filter query parameter. 
curl "http://localhost:8090/api/projects/default/envs/prod/collections/posts?filter=%7B%22op%22%3A%22eq%22%2C%22path%22%3A%22%24.data.status%22%2C%22value%22%3A%22published%22%7D" \
  -H "Authorization: Bearer $SILO_KEY"
A filter may test at most 16 leaf conditions. Filters with more leaves are rejected with 400 validation_failed. Break large filters into multiple requests if needed.

​
Sorting

The sort parameter accepts a comma-separated list of JSONPath expressions. Prefix a path with - for descending order. 
?sort=-$.updated_at,$.data.title
  • Sort paths must select at most one node per entry. Array paths ([*]) are not valid sort keys.
  • Multiple sort terms are applied left to right.
  • When no sort is given on a search request, results are ranked by relevance.

​
Pagination

?limit=50&offset=0
ParameterDefaultMaximum
limit50500
offset0—

​
Response envelope

Every list and search response wraps results in a consistent envelope: 
{
  "data": [...],
  "total": 137,
  "limit": 50,
  "offset": 0
}
total is the count of all matching entries, regardless of the current page. Use it to drive pagination UI.

​
Full-text search

Search takes the same filter, sort, limit, and offset parameters, plus q for the search text. It is available at three reaches:
ReachPath
One collection/api/projects/{project}/envs/{env}/collections/{name}/search?q=…
One environment/api/projects/{project}/envs/{env}/search?q=…
Everything the key can read/api/search?q=…
The reach is determined by the path, never by a parameter, so a typo cannot accidentally widen a search to a scope you did not intend.
Which fields are full-text indexed is a per-collection schema decision, controlled by the x-silo-search keyword. An anonymous caller can only reach collections whose schema does not set x-silo-auth.

​
TypeScript client

The @org-quicko/silo-client package ships a type-safe filter builder. The result is the same JSON AST described above. 
import { Filter, Sort } from "@org-quicko/silo-client"

const page = await posts.list({
  where: posts.filter.field("status").equals("published")
    .and(posts.filter.each("tags").equals("release")),
  sort: Sort.recentlyUpdated(),
  limit: 50,
})
posts.filter.field("status") addresses $.data.status. posts.filter.each("tags") addresses $.data.tags[*]. Sort.recentlyUpdated() expands to sort=-$.updated_at.

Build docs developers (and LLMs) love

Get started for freeTalk to us