Skip to content

gooseworks-ai/goose-skills / skills/marketing/work-with-creators

work-with-creators

Verified

Real creators and influencers, researched, priced and briefed. Use when the marketer says "find influencers", "which creators are worth paying, at what price?", "find real UGC creators to film for our ads", "who should we seed product to", "is this creator's engagement real", "what should I pay this creator" or "write a creator brief". Free research first; one credit yes only to vet deeper. Drafts only: contacts nobody, agrees nothing, moves no money. Not for AI-made UGC-style video — use goose-video; if unsure, ask: a real person filming, or AI-made?

references/run-memory.md

View on GitHub
# Run memory for work-with-creators

How this playbook uses [the `run-memory` skill](../../run-memory/SKILL.md): Start before discovery, End after the report. Only what is specific here.

## Jobs

| `job` | `subject` | `cadence` | Runs when |
|---|---|---|---|
| `creator programme` | `wave-<n>` — the wave this run plans or reads | `ad hoc`, or `monthly` when a roster refresh was set up | The shortlist that proposes wave *n*, a refresh or re-vet while wave *n* is open, and the read of wave *n*'s results — all under the same subject |

**One job, one subject per wave, from proposal to results.** The shortlist run that recommends wave 1 saves its record under `creator programme` / `wave-1`; the results read of wave 1 runs under the same job and subject, finds that record directly, and judges the wave-1 recommendation **by its original id**. A run that proposes the next wave opens `wave-2` (and first reads the newest `wave-1` record to carry its open recommendations forward with their ids). A first run with no wave yet uses `wave-1`.

## Start: "since last time"

**Outcome sources:**
- **The roster** (`creators/roster.csv`, outcomes-ledger `entity: creator`) — the per-creator truth. Recount from it: shortlisted, contacted, replied, quoted, declined, signed, posted — and for each wave, `wave_result` against its criterion (`win · partial · miss · no-data`), grouped by role, tier, deal structure and vetting depth.
- **Wave files** (`creators/<D>/wave-<n>.md`): each wave's criteria and read window — is the window over, and were the results reported?
- **Lessons** (`creators/lessons/`) and their proposed changes.
- **The brain**, with `brand_knowledge_*` tools: `brand_knowledge_search` since the last record, `source_types: ["learning", "creative", "meta_ad"]` — new `[creators …]` rules, and creator content turned into ads since (chapter 22, "Everything we have made and run"), so a creator whose content now runs as an ad is visible. Otherwise `knowledge_search`. `campaign_read` for a campaign's creator leg.

**Check last run's recommendations:** "send wave 1 to these 8" is `done` when the roster shows them `contacted`; "re-vet after their next 10 posts" when the roster's `revet_by` passed and a vet ran; a price band is judged against the quotes that came back (`quoted_rate` vs the recommended range). Verdicts only from what the roster shows — reported by the founder.

**Say, in 3–5 lines:** e.g. "Since 2 Oct: wave 1 went to 8 creators; 6 posted, 3 hit the views target (2 nano, 1 micro), the 2 gifted-only picks didn't post. Quotes came back inside our range for 4 of 5. Two creators' content now runs as ads." Then use it: renew the wins under their renew rule, cut the misses, shift the mix toward the role and tier that worked — with the sample-size rule (under ~5 creators in a group → "directional only").

## End: the record

- **Numbers:** creators found, vetted (public vs paid depth), shortlisted, by role and tier; price bands proposed; for a wave read — sent, posted, `win / partial / miss / no-data` counts by role, tier and deal structure, quotes vs range; credits spent. Each with its window and label (`reported` for the founder's outcomes, `derived` for our counts, `public` for public post numbers).
- **Recommendations:** wave 1 (or the next wave) with its criteria, renew / cut calls, price bands — each with an id, owner (`marketer` — the founder sends and pays), review-by (the wave's read window end) and how we'll know (the wave criterion's metric and threshold, from `wave-<n>.md`).
- **Links:** `Report:` the report; in "Checked", the roster, the wave file, the lessons file.

## Never in a run record

No creator handle, name, rate quoted by a named creator, email or profile link — groups, counts and the wave id only. The roster holds the creators; the record links it. (Public numbers about a named creator live in the report and the roster, never in the record.)
add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall