gooseworks-ai/goose-skills / skills/marketing/research-competitors
research-competitors
VerifiedCompetitor research: what every main rival is running, launching, charging and saying now, what works for them, and what that means for you. Use when the marketer asks for competitor research or a teardown, "what ads are my competitors running", "how do they price", "find gaps and angles", "what changed this week" (a weekly watch), or "X just launched / shut down / cut prices" (react now). Also builds a sales battlecard, checks a competitor page's tools/pixels, or studies who engages with their LinkedIn posts. Delivers a readable report saved to the brand, so the next run reports what changed. Never copies their words or disparages them. Not for what people say about you or your market — use research-market; not for who your buyers are — use research-audience.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 11.7 KB |
| references/angles-and-gaps.md | markdown | 7.4 KB |
| references/battlecard.md | markdown | 3.8 KB |
| references/focused-outputs.md | markdown | 2.7 KB |
| references/linkedin-engagers.md | markdown | 6.3 KB |
| references/page-technology.md | markdown | 2.7 KB |
| references/reading-ads-honestly.md | markdown | 8.6 KB |
| references/report-format.md | markdown | 10.0 KB |
| references/research-spec.md | markdown | 34.5 KB |
| references/rival-craft.md | markdown | 16.0 KB |
| references/run-memory.md | markdown | 6.6 KB |
| references/watch-and-react.md | markdown | 12.7 KB |
| skill.meta.json | config | 5.5 KB |
| tests/examples/2026-10-22/battlecard.md | markdown | 5.5 KB |
| tests/examples/2026-10-22/engagement-report.md | markdown | 3.7 KB |
| tests/examples/2026-10-22/engagers.csv | other | 1.7 KB |
| tests/examples/2026-10-22/excluded.csv | other | 1.0 KB |
| tests/examples/2026-10-22/technology.md | markdown | 3.2 KB |
| tests/fixtures/engagers.csv | other | 1.9 KB |
| tests/fixtures/focused-evidence.md | markdown | 3.8 KB |
| tests/fixtures/page-source.html | other | 614 B |
| tests/test-by-hand.md | markdown | 7.5 KB |
references/linkedin-engagers.md
View on GitHub# Human version
This finds evidence of who comments on or reacts to a competitor's LinkedIn posts. It keeps the source post and engagement type, checks coverage, and separates participation from buying intent. A supplied export works even when a managed collection route is unavailable.
---
# Agent version
## Source and scope
1. Take company/post URLs, requested window, post count and audience filters from the ask/brain. Defaults when absent: last 30 days, top 3 recent posts by reactions plus comments; state these assumptions. With supplied post URLs, inspect those posts rather than substituting popular ones. The `marketing-data-access` skill (`../../marketing-data-access/SKILL.md`) and the `spend-gate` skill (`../../spend-gate/SKILL.md`) govern managed reads and spending. Post search/feed/profile access does not imply commenter/reactor access.
2. **Managed access when listed:** use `gooseworks:marketing_data_describe` to check `linkedin_comments` / `linkedin_reactions`; its live schema is the contract. The matching `gooseworks:marketing_data_lookup` quote is free; execute only after the user approves that quote and ceiling, following marketing-data-access. Read the actual response before mapping person fields; a schema saying only `results[]` is not a per-person field guarantee. These tools are conditional: an older connector without them uses the next paragraph and reports the managed gap.
Choose an available evidence source, in this order: the user's export; a **verified live managed route** whose documented response includes per-person commenter/reactor rows and source post linkage; public professional comments the host can actually read. Keep collection method and retrieval date. Verify route/schema before spending; use only the listed provider and documented keys, within an approved ceiling. No customer vendor key, unofficial API, login workaround or invented endpoint. A public sample cannot promise all reactors. If no source yields person-level rows, return the aggregate post inventory plus the exact missing export/managed operation; mark the extraction incomplete.
3. For an account rather than supplied posts: retrieve its posts in the window, dedupe by canonical post URL, rank by the stated metric, select the requested count, and retain numerator/coverage (posts retrieved, in-window, selected; any paging cap). Inspect engager pages/rows for only those selected posts. Without complete feed coverage say “top among the posts retrieved”, never the company's definitive top posts. If metrics are unavailable, retain supplied order and say ranking was not possible.
## Keep evidence, then qualify
4. Normalise to private rows: `profile_url, name, headline, company, source_post_url, post_date, engagement_type, comment_text, engagement_date, retrieved_date, method, evidence_status, fit_reason`. A valid evidence row needs a canonical person profile URL, the selected source post and a supported type (`comment` or `reaction`). Drop unsupported rows to `excluded.csv` with a reason. Preserve unknown dates/fields explicitly. Do not claim an engagement occurred this month from the post date alone: state whether the window covers **posts published** or **engagements dated**, and flag unknown engagement dates. Discard unrelated company posts and rows outside the requested kind/window.
5. Dedupe people by profile URL, never by name; preserve every source-post/type pair and each actual comment. Keep “reacted” separate from what the comment says. Do not infer pain, product use, intent or employer from a like. Exclude the rival's or seller's staff only when a public current source or the export identifies that relationship; ambiguous affiliation stays unknown. For an audience filter, show supported match / not a match / unknown with the actual role/company evidence, not a title-keyword guess presented as qualified.
6. **Qualified prospects:** this output is engagement evidence. When the user also asks for prospects, use the existing audience list workflow on these supplied evidence rows (the human's request authorizes that additional job); its account/people fit, exclusions, suppression and contact checks still apply. The current audience rule excludes sourcing people merely because they liked or replied around a pain post. Do not silently override it: use an authored professional statement of the relevant pain or other independently supported fit/signal, or deliver the engagement rows only and state which qualification remains unavailable. A collection route does not settle the prospecting rule.
## Deliver and protect the evidence
- `research/competitors/engagers/<rival>/<date>-engagers.csv`: one row per unique person/post/type; include the schema above. `excluded.csv` beside it holds dropped rows/reasons. Workspace only, under the brand's scope; use the host's CSV writer or workspace file tools, valid UTF-8 quoting. Formula-like text (`=`, `+`, `-`, `@`, tab or carriage return at the start) is escaped for spreadsheet viewing. Preserve the original comment in the private source export so quoting cannot be mistaken for changed evidence.
- The focused **report** gives the window definition, selected posts with links, ranked-post coverage, raw rows → kept evidence rows → unique people, counts by engagement type, aggregate role/company themes, fit unknowns and data gaps. Link the private CSV for the user; names, profile URLs and comment text stay out of brain-imported reports, learning facts, shared price logs and run records. Run records hold counts and source-post URLs only.
- Save the aggregate report via focused-outputs.md; write private row files directly with `gooseworks:file_write` in the brand workspace, never knowledge_import. No contacting anyone or claims of consent to outreach.
## Completion check
Every kept row traces to a selected post and a source record. Two people with the same name stay distinct; the same person across posts stays linked. Counts reconcile with exclusions and duplicates; unknown engagement dates do not pass an engagement-date filter. The report states exact coverage and whether managed collection, an export or a public sample supplied it. A managed extraction is only complete after an actual route succeeds; a supplied-export run proves export processing only. Missing managed collection remains an explicit blocker in the report, not a successful empty list.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |
| 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: 115k | Updated | Claude CodeCodex | Verified | Install |