Skip to content

gooseworks-ai/goose-skills / skills/marketing/research-market-edge-cases

research-market-edge-cases

Verified

Step of the research-market playbook: rules for run edge cases such as no browser, stalled workers, earlier reports and late reads. Run research-market instead; not for direct use.

FileKindSize
SKILL.mdskill md2.3 KB
skill.meta.jsonconfig244 B
---
name: research-market-edge-cases
description: >-
  Step of the research-market playbook: rules for run edge cases such as no browser, stalled workers,
  earlier reports and late reads. Run research-market instead; not for direct use.
version: 0.1.1
---

# edge-cases

- **No browser at all** (hosted chat, a scheduled run): use search results, the pages fetch can read, and the free server reads (`brand_read` `organic`, `gooseworks:social_inspiration_search`). Say once, in the report's "In short", what couldn't be read and what it means for the answer; the paid top-up for it is a decision in the closing message, never a question before the report.
- **The browser waits on the founder's approval, or keeps asking:** tell them in one line what it is for, then finish the same reads with search and fetch (`site:` queries, each post's own link) — first-hand evidence such as the brand's own users is finished another way, never dropped. List only what still couldn't be read. Never sit silent.
- **A worker stalls** for more than a few minutes: say so in one line. **A worker is stopped:** keep everything it returned or wrote; never throw read work away.
- **Facts from earlier saved research** (dates, figures, launches): re-check them, or mark them "from earlier research, not re-checked".
- **A saved rival that isn't one** (a person's account, a namesake, a dead company): drop it, and name it in the report so the founder can fix the list.
- **An earlier report on this brand exists:** decide on purpose, and say which in one line — in chat and in the report's first "In short" bullet. Newer than 30 days (a monthly pulse covers less) and the ask doesn't say "fresh" or "update": **reuse** it as the base — re-check only what moves (news, launches, prices, the conversation) and deliver a what-changed report; never re-deliver it untouched. Older, or the founder asks for a fresh look: **refresh** in full. A report from today is replaced at the same path. Never treat a saved report's facts as checked without re-checking the ones the new report leads with.
- **A read lands after the report:** rewrite the whole report, top lines included. Never patch a section in.
- **Status notes are not questions:** one-line notes (the browser heads-up, a stalled worker, a serious risk) are fine before the report; questions are not.
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