Output formats & schema - Bron Developer

Every command supports four output formats and a small projection toolbox:

--output json (default) — pretty-printed JSON, colored on a TTY.
--output yaml — YAML, useful for hand-editing.
--output jsonl — one JSON object per line, ideal for streaming and jq.
--output table — fixed-width ASCII table.

Plus orthogonal modifiers:

--columns a,b,c — keep only the listed dot-paths (works for json / yaml / jsonl / table).
--cell-max <N> — truncate table cells; default 28, 0 disables truncation entirely.
--embed <token> — pull related entities/values into _embedded per spec-defined token.
--schema — emit the JSON Schema for the command instead of running it.

Choosing a format

bron tx list --output json     # default
bron tx list --output yaml
bron tx list --output jsonl    # one record per line — pipe-friendly
bron tx list --output table    # ASCII grid, *At fields render as ISO UTC, nested objects collapse to {…}/[…N]

json is colored when stdout is a TTY. Disable with NO_COLOR=1, force on with FORCE_COLOR=1.

Projecting fields with `--columns`

--columns accepts a comma-separated list of dot-paths. The command keeps only those keys, in the listed order. Works for every format:

# Table view of just the columns you care about.
bron tx list --output table \
  --columns transactionId,status,transactionType,createdAt

# Same projection, but as JSON.
bron tx list --output json \
  --columns transactionId,status,params.amount

# Dot-paths drill into nested objects (table flattens, json/yaml/jsonl emit nested).
bron tx list --output table \
  --columns transactionId,params.amount,params.assetId,_embedded.asset.symbol

For a list endpoint, --columns projects each item in the list (e.g. each transaction). For a single resource it projects the top-level object. A dot-path that crosses an array applies to every element — --columns transactionId,_embedded.events.usdAmount keeps usdAmount on each event of each transaction.

In the MCP server, the same projection is the fields tool argument, and heavier transforms are the jq argument — applied server-side so the trimmed result is all the agent’s context ever sees. There’s no shell to pipe through, so the shaping moves into the call itself.

Filtering and transforming — pipe to `jq`

The CLI keeps output shaping minimal on purpose: --columns for projection, --output jsonl for line-based piping, and then jq for everything heavier (select, slicing, arithmetic, joins). One mental model, one tool to learn:

bron tx list --output jsonl \
  | jq 'select(.status == "signed") | .transactionId'

bron balances list --embed prices --output jsonl \
  | jq -s 'map(._embedded.usdValue | tonumber) | add'

Date and number handling

Decimals stay strings end-to-end ("0.0000123") — no float64 rounding, no precision loss. Don’t unmarshal them as Number in your scripts.
Date fields render as ISO 8601 UTC in every format (2026-04-30T11:14:13.274Z), regardless of how the spec encodes them on the wire (epoch millis vs ISO).
Date-shaped query parameters (names ending in AtFrom, AtTo, Since, Before, After) accept both ISO 8601 (2026-04-01T00:00:00Z, 2026-04-01) and millisecond-epoch integers — the CLI normalises to millis before sending.

Some list endpoints expose --embed <token> to fold related data into the response under _embedded, saving you a follow-up query. The available tokens depend on the resource:

# Add USD prices and computed values to every balance.
bron balances list --embed prices \
  --output table --columns symbol,totalBalance,_embedded.usdPrice,_embedded.usdValue

# Resolve assetId on each transaction into _embedded.asset.
bron tx list --embed assets \
  --output table --columns transactionId,_embedded.asset.symbol,params.amount,status

Token discovery: bron <r> <v> --help lists the supported --embed values for that command. Internally each token routes to the matching backend includeXxx query parameter. _embedded follows the HATEOAS pattern — it’s an opt-in extra layer on top of the spec-defined response shape. A consumer that didn’t ask for an embed never sees the extra keys, so generated SDK types stay aligned with the published schema.

In --schema the field is named embedded, but it serialises as _embedded on the wire — so a --columns (or MCP fields / jq) path must use the underscore: _embedded.usdValue, not embedded.usdValue. The MCP bron_help tool already prints the wire name.

`--schema` for machine consumers

--schema emits the JSON Schema for any command — request body, query params, response shape, every documented status code:

bron --schema                            # full CLI as one OpenAPI 3.1 document
bron <resource> <verb> --schema          # single-command fragment
bron help --schema                       # alias for the full schema dump
bron --output yaml <r> <v> --schema      # same dump in YAML

--schema is strict OpenAPI 3.1. Pipe it to jq, swagger-cli, or any spec consumer. For LLM agents this is the canonical entry point: one call to discover every command, every flag’s type and enum values, and every response field. See LLM agent integration for the full agent flow.

Pipelines with `jq`

JSONL is the easiest format for shell pipelines. Combine it with jq for ad-hoc transformations:

# Count transactions by status over the last day.
bron tx list --createdAtFrom "$(date -u -v-1d +%Y-%m-%dT%H:%M:%SZ)" --output jsonl \
  | jq -r '.status' | sort | uniq -c

# Pluck transaction IDs of every withdrawal awaiting approval.
bron tx list --transactionStatuses waiting-approval --transactionTypes withdrawal --output jsonl \
  | jq -r '.transactionId'

# Stream live updates and react in shell.
bron tx subscribe --transactionStatuses signing-required \
  | jq -r 'select(.transactionType == "withdrawal") | .transactionId' \
  | xargs -I {} bron tx approve {}

The CLI also accepts machine-readable input on stdin where it makes sense — bron tx withdrawal --file - reads a JSON body from stdin, useful for scripting or piping from another tool.

​Choosing a format

​Projecting fields with --columns

​Filtering and transforming — pipe to jq

​Date and number handling

​--embed for related entities

​--schema for machine consumers

​Pipelines with jq

Choosing a format

Projecting fields with `--columns`

Filtering and transforming — pipe to `jq`

Date and number handling

`--embed` for related entities

`--schema` for machine consumers

Pipelines with `jq`