Menu
Run a small sample through Clay enrichment columns and verify the results - use when you need to fire an action on rows (credit-consuming), watch a run to completion, or read back and score a table before deciding whether to scale.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Set up auth, session, and profile config for clay-cli before running any command - use when a Clay command fails with "session expired", when starting fresh in a new checkout, or when configuring workspace/folder/workbook IDs for the first time.
Discover Clay actions and manage the local action catalog - use when searching for an available action/integration, checking catalog coverage against the strict proof registry, classifying action safety, or generating action templates.
Extract Clay table/workbook data to local files - use when you need a full-fidelity workbook export, a template-shaped snapshot of a table, or raw paginated row dumps for offline processing.
Run a full Clay playbook end to end - intake, plan, prepare, sample, verify, scale - use when a plain-language request should be routed to one of the 12 packaged playbooks and operated through to a safe scale decision, not just a single ad-hoc command.
Build Clay list-building sources - company-to-people waterfalls, source preview/import, and webhook intake - use when adding a new list-building source to a table or converting a companies table into a people search.
Build and modify Clay tables, views, fields, and records with clay-v2.js - use when creating a new table/workbook, adding or editing columns (text/formula/http-api/use-ai/action), managing views, or writing declarative table specs.
| name | clay-run-enrichment |
| description | Run a small sample through Clay enrichment columns and verify the results - use when you need to fire an action on rows (credit-consuming), watch a run to completion, or read back and score a table before deciding whether to scale. |
run-top spends Clay credits - it is the only command in this package in the
CREDIT set as well as MUTATING. Always sample small, always read back before
deciding to scale.
node clay-v2.js run-top t_TEST_TABLE --field f_TEST_FIELD --view gv_TEST_VIEW --n 10 --confirm
--n defaults to 10 and the CLI refuses anything above 10 unless you also pass
--allow-more-than-10. Do not reach for that flag to skip the sample-first workflow
node clay-v2.js run-watch t_TEST_TABLE --field f_TEST_FIELD --timeout 300 --interval 5
node clay-v2.js run-status t_TEST_TABLE --workspace {{workspace_id}}
run-watch polls until the field finishes or the timeout elapses (default 300s,
poll every 5s by default). Both are read-only.node clay-v2.js verify-table t_TEST_TABLE --view gv_TEST_VIEW --include-rows 10 --require-values f_email,f_status
node clay-v2.js proof-readback t_TEST_TABLE --view gv_TEST_VIEW --field f_TEST_FIELD --include-rows 10
verify-table --require-values fails if any of the named fields are empty across
the sampled rows - use it as your first readback gate after a run-top.proof-readback is hard-capped at 10 rows (--include-rows cannot exceed it) and
accepts --expected-enums-json '{...}' to check enum-typed outputs against an
expected set. It also supports a fully offline mode via --from-manifest manifest.json if you already pulled a manifest with /clay-explore.node clay-v2.js redact manifest.json --report redact-report.json
node clay-v2.js score before-manifest.json after-manifest.json
redact strips sensitive keys/values from any exported JSON down to
<redacted:...> stubs, useful before sharing a manifest for review.score compares field-config parity between two manifests (e.g. a spec's expected
shape vs the live table) - both are fully offline, no session required.For anything beyond a one-off sample, /clay-run-playbook wraps this exact
sample -> readback -> evidence -> quality-report -> scale-gate sequence with explicit
confirmation gates at each step. Reach for this skill directly when you just need one
sample run and a readback; reach for /clay-run-playbook when you're running a
full playbook end to end.
run-top and run-status/run-watch all need a valid session - see
/clay-onboarding if you get a session-expired error mid-run.verify-table and proof-readback are read-only but still Clay-scoped: they need
the same session as everything else, they just never mutate.