@zoominfo/gtm-ai-cli 2.1.0 → 2.2.0

package/README.md

@@ -2,7 +2,8 @@
 
 A command-line interface for searching [ZoomInfo](https://www.zoominfo.com/) go-to-market (GTM) data. 
 Search and enrich companies and contacts, surface intent / news / scoop signals, browse engagement history, 
-manage GTM context, and pipe results through your shell as JSON, JSONL, CSV, YAML, or a table.
+manage GTM context, clean and score your own records, send records to connected systems, and pipe
+results through your shell as JSON, JSONL, CSV, YAML, or a table.
 
 ## Installation
 
@@ -87,20 +88,20 @@ gtm auth login
 
 ## Output formats
 
-Every subcommand accepts `-f, --format`. The `table` and `csv` formats auto-flatten JSON:API envelopes (`data: [...]`) and one-level-nested objects (`company.id`, `company.name`); `json`, `jsonl`, and `yaml` preserve the exact response shape.
+Every subcommand accepts `-f, --format`. The `table` and `csv` formats auto-flatten JSON:API envelopes (`data: [...]`) and one-level-nested objects (`company.id`, `company.name`); `jsonl` emits each record of a list response on its own line; `json` and `yaml` preserve the exact response shape.
 
 | Format | When to use |
 |---|---|
 | `json` (default) | Pretty-printed JSON. Pipe to `jq` for field extraction. |
 | `jsonl` | One JSON object per line. Stream into log/data pipelines. |
-| `csv` | Flat CSV with headers; nested objects are dot-flattened, arrays are stringified. |
+| `csv` | Flat CSV with headers; nested objects are dot-flattened, arrays are stringified. Text cells that a spreadsheet would run as a formula (leading `=`, `+`, `-`, `@`) are prefixed with `'`. |
 | `yaml` | Human-readable diffs of deeply nested responses. |
 | `table` | ASCII bordered table. Best for terminal browsing of small responses. |
 
 ```shell
 gtm companies search --industry software -f table
 gtm contacts search --management-level "C Level Exec" -f jsonl > c-level-contacts.jsonl
-gtm-context get -f yaml
+gtm gtm-context get -f yaml
 ```
 
 ### Projecting fields with `--select`
@@ -346,16 +347,16 @@ This is the context that shapes how every other ZoomInfo tool interprets your qu
 
 ```shell
 # Compressed view (~1k tokens)
-gtm-context get
+gtm gtm-context get
 
 # Full expanded profiles
-gtm-context get --detailed -f yaml
+gtm gtm-context get --detailed -f yaml
 
 # Update — synthesizes unstructured content into structured records (admin only)
-gtm-context update --query "Update competitive intelligence" --source "Our main competitors are X, Y, Z. X is strong in …"
+gtm gtm-context update --query "Update competitive intelligence" --source "Our main competitors are X, Y, Z. X is strong in …"
 
 # From a file (large inputs)
-gtm-context update --query "Create buyer personas from these interviews" --source-file ./interview-notes.txt
+gtm gtm-context update --query "Create buyer personas from these interviews" --source-file ./interview-notes.txt
 ```
 
 ---
@@ -470,6 +471,69 @@ call). Column and row IDs come from `gtm audiences get`.
 
 ---
 
+### `gtm data-quality` (alias `dq`) — prepare, segment, and score your own records
+
+Data Quality (beta) cleans and classifies records you bring and returns the results inline;
+nothing is stored or written to another system. The default prepare steps (parse, validate,
+normalize) are free. `--line-type`, `--verify`, custom normalization (`--config-id`, `--rules`),
+`score`, and `segment` use credits.
+
+```shell
+# Clean emails, phones, addresses, job titles, or person names (up to 30 per call)
+gtm dq prepare email " Jane.Doe@Example.COM " "jane@" --select input.email,value.fullEmail,isValid -f table
+gtm dq prepare phone "(617) 555-0123" --line-type
+gtm dq prepare address --file ./addresses.json --verify
+
+# Custom normalization: a saved config, or inline rules grouped by rule group
+gtm dq prepare phone "+1 617 555 0123" --config-id <configId>
+gtm dq prepare phone "+1 617 555 0123" --rules ./phone-rules.json   # { "phone": [{ "type": "INT_FORMAT", "value": "E164" }] }
+
+# Segment or score up to 25 records against saved configs
+gtm dq segment --config-ids <configId> --file ./leads.json
+gtm dq score --config-ids <configId> <configId> --file ./leads.json
+
+# Results are in input order; e.g. each lead's email and first score as CSV
+gtm dq score --config-ids <configId> --file ./leads.json \
+  | jq -r '.data.attributes.records[] | [.input.email, .outputs[0].value] | @csv'
+
+# Manage configs (kinds: normalize, score, segment)
+gtm dq configs list score
+gtm dq configs get score --id <configId>
+gtm dq configs create score --file ./icp-score.json
+gtm dq configs update score --id <configId> --file ./icp-score.json
+gtm dq configs delete score --id <configId>
+```
+
+`prepare --file` takes a JSON array of raw strings, or of objects holding pre-parsed components
+(e.g. `{ "number": "6175550123", "countryCode": "US" }` for a phone) to skip parsing. Results are
+positional: `data[i]` is the result for value `i`, and a value that can't be processed returns a
+`PrepareError` in its slot. `score` / `segment --file` takes a JSON array of records keyed by the
+field names the configs use. Config files hold the config's attributes (for `update`, send the
+complete config for score and segment configs); see the
+[Data Quality API reference](https://docs.gtm.ai/reference/preparecontroller_prepare) for each
+kind's schema.
+
+---
+
+### `gtm exports` — send records to connected systems
+
+Exports (beta) send ZoomInfo companies or contacts to the systems your organization has connected,
+such as a CRM. Running an export writes to that system; runs are asynchronous.
+
+```shell
+gtm exports list -f table
+gtm exports get --id <exportId>
+
+# Send up to 50 records as one run, then check on it
+gtm exports run --id <exportId> --company-ids 344589814 239305146
+gtm exports status --id <runId>
+
+# Or wait for the run to finish (exits non-zero if it fails or is stopped)
+gtm exports run --id <exportId> --contact-ids 1260398587 --wait
+```
+
+---
+
 ### `gtm raw` — universal escape hatch
 
 Use this when you need a tool that doesn't have a curated wrapper yet.