{"id":19527,"plugin_id":"plugins_6a9a2e84c7488191b7586c5593b7ca1a","kind":"skill","collection_source":null,"comparison_source":null,"observed_at":"2026-09-30T23:15:38.121Z","digest":"14e94c9b9ec871bfa822d958a08c6760634d5e4da0acdd9f835348867e8f42a3","against":null,"payload":{"description":"Use when the user asks to use the Freckle CLI or get Freckle to do something: enrich/score/process a lead list, build a company or people list, import CRM records, monitor Dataset Signals, build or change Workbooks, datasets, or Workflows, publish or run saved workflows, inspect credit usage or node capabilities, auth/org/config setup, or integration connections.","included_files":[{"relative_path":"BUILD.md","size_in_bytes":6893},{"relative_path":"CONNECTIONS.md","size_in_bytes":9083},{"relative_path":"CREDITS.md","size_in_bytes":6494},{"relative_path":"LIST.md","size_in_bytes":13858},{"relative_path":"REFINE.md","size_in_bytes":5718},{"relative_path":"SETUP.md","size_in_bytes":2895},{"relative_path":"SIGNALS.md","size_in_bytes":9054},{"relative_path":"WHAT-CAN-FRECKLE-DO.md","size_in_bytes":3392},{"relative_path":"WORKBOOKS.md","size_in_bytes":18803},{"relative_path":"list/ai-ark-companies.md","size_in_bytes":463},{"relative_path":"list/ai-ark-people.md","size_in_bytes":532},{"relative_path":"list/apollo-companies.md","size_in_bytes":1060},{"relative_path":"list/apollo-people.md","size_in_bytes":1691},{"relative_path":"steps/step-1-context.md","size_in_bytes":2370},{"relative_path":"steps/step-2-objective.md","size_in_bytes":6755},{"relative_path":"steps/step-3-contract-map.md","size_in_bytes":5944},{"relative_path":"steps/step-4-plan.md","size_in_bytes":5713},{"relative_path":"steps/step-5-draft.md","size_in_bytes":6157},{"relative_path":"steps/step-6-run.md","size_in_bytes":6472},{"relative_path":"workflow/apollo-find-people.md","size_in_bytes":5945},{"relative_path":"workflow/cli-reference.md","size_in_bytes":11390},{"relative_path":"workflow/collector.md","size_in_bytes":6815},{"relative_path":"workflow/credit-cost.md","size_in_bytes":7041},{"relative_path":"workflow/draft-syntax.md","size_in_bytes":9485},{"relative_path":"workflow/push-to-dataset.md","size_in_bytes":4283},{"relative_path":"workflow/research-agent.md","size_in_bytes":4096},{"relative_path":"workflow/waterfall.md","size_in_bytes":2437}],"name":"freckle","skill_md_contents":"---\nname: freckle\ndescription: \"Use when the user asks to use the Freckle CLI or get Freckle to do something: enrich/score/process a lead list, build a company or people list, import CRM records, monitor Dataset Signals, build or change Workbooks, datasets, or Workflows, publish or run saved workflows, inspect credit usage or node capabilities, auth/org/config setup, or integration connections.\"\n---\n\n# Freckle\n\nThis is a router skill: pick one primary route below and load adjunct references only when that route requires them. Load the route's file before running any `freckle` command — even `--help`; the route files tell you which commands to run and when.\n\n**New builds are gated.** Route work with no existing target to [BUILD.md](BUILD.md). When the request identifies an existing Workbook or Workflow by URL, id, or name, route to [REFINE.md](REFINE.md) instead: locate the target, derive its org, and work from the design already embodied there. REFINE owns the fit test and escalates only a genuine redesign or a shape-changing ambiguity.\n\n## Route\n\n| User wants | Load |\n| --- | --- |\n| List building — build a new list of companies or people from a provider search, or run/inspect a List Workbook created by the `list` commands | [LIST.md](LIST.md), then selected command help; provider references on demand |\n| Import a CSV or plain rows into a **new** Workbook, with no enrichment or workflow asked for and no existing target named | [BUILD.md#data-import-fast-lane](BUILD.md#data-import-fast-lane) — a pure import that names an existing Workbook or Dataset routes to [REFINE.md](REFINE.md) instead |\n| Get Freckle to do something with no existing target — enrich, score, look up, run data, or build a Workbook or Workflow | [BUILD.md](BUILD.md) |\n| Do anything to an existing Workbook or Workflow identified by URL, id, or name — add data, run rows, or change its setup | [REFINE.md](REFINE.md) |\n| Run an existing saved Workflow directly with JSON inputs, inspect runs, list saved Workflows | [workflow/cli-reference.md](workflow/cli-reference.md) |\n| Inspect node capabilities | [workflow/cli-reference.md#node-catalog](workflow/cli-reference.md#node-catalog) |\n| Workbook housekeeping that changes nothing about what runs — inspect, archive, delete entries, rotate webhook secrets | [WORKBOOKS.md](WORKBOOKS.md) |\n| Return a Dataset webhook endpoint URL | [WORKBOOKS.md](WORKBOOKS.md) |\n| Create, inspect, list, turn monitoring on or off, or delete Dataset Signals | [SIGNALS.md](SIGNALS.md) |\n| Check the current credit balance or report credit usage by Workbook, Workflow, or billed node | [CREDITS.md](CREDITS.md) |\n| Log in, check auth, choose org, inspect endpoints | [SETUP.md](SETUP.md) |\n| Connect Apify, ContactOut, HeyReach, HubSpot, Instantly, Lemlist, OpenAI, Salesforce, Slack, Supabase, or Twain, or find credential IDs | [CONNECTIONS.md](CONNECTIONS.md) |\n| Set up, list, or pick a credential so Workflows can use the user's own API key — match supported provider BYOK first, then custom HTTP APIs; a build or change request that mentions their own key still routes to BUILD or REFINE | [CONNECTIONS.md](CONNECTIONS.md) |\n| Understand enrichment waterfalls, collectors, or Research Agent usage | [workflow/waterfall.md](workflow/waterfall.md), [workflow/collector.md](workflow/collector.md), [workflow/research-agent.md](workflow/research-agent.md) |\n| Explain what Freckle can currently do — a capability question only, with no task attached | [WHAT-CAN-FRECKLE-DO.md](WHAT-CAN-FRECKLE-DO.md) |\n\n## Shared Operating Rules\n\n- When user decisions are needed, run a **batch grill**. Map the decisions as a design tree and work it in rounds. The frontier is every decision whose prerequisites are settled: ask the whole frontier in one round, number every question, and give your recommended answer for each, then wait for the user's answers. A decision that depends on another unsettled answer waits for a later round. After each reply, recompute the frontier. The grill is complete only when the frontier is empty: every shape-changing branch has been visited and nothing remains silently assumed. Make changes or run data only after the user confirms shared understanding; a route's later approval gate may supply that confirmation.\n- Every question round and every plan opens with a **what-will-run summary**: two or three plain sentences saying what Freckle will do once the user says go — which rows it reads, what happens to each, where results land, and what it costs — in product words only, with no flag names, node keys, or field paths. The tables, diagrams, and commands that follow are evidence for the summary, never a substitute for it.\n- Facts are agent work. Resolve anything available from the CLI, filesystem, inspected artifacts, catalogs, or other tools instead of asking the user. If a lookup can run independently, dispatch it and ask the rest of the frontier while it runs; only decisions downstream of that lookup wait.\n- Names are agent work too. Choose the label for every new Workbook, Dataset, and Workflow yourself — short, descriptive, in the user's own words for the goal — and show it wherever the user meets the artifact: the what-will-run summary, the plan, and the final link. Use a name the user supplies verbatim; otherwise the user renames in the app if they prefer something else.\n- Resolve the org from any named resource before asking the user. Confirm the active org only when the request supplies no resource from which to derive it. Auth-only, config-only, and generic product-capability answers need no org setup.\n- After automatic resolution, user selection, or an explicitly supplied organization, append `--org-id=<org-id>` after the complete subcommand path of every subsequent CLI command — for example, `freckle workbook list --org-id=<org-id>` — preserving any organization the user supplied. Do not use `freckle org switch`; it writes shared global config that another agent can overwrite.\n- Preflight before the first command: `command -v freckle && freckle whoami`.\n- `--json` always means output: structured results render YAML by default and pretty JSON with `--json`. Commands that support tables require `--table`. Generic inline JSON payloads use `--input-json`; command help documents specialized JSON filters and file inputs.\n- Confirm unfamiliar flags with `freckle <subcommand> --help` before running a command your route file prescribes.\n- Use absolute `--file` paths when possible; package scripts can change relative resolution.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}