Lightster (lightster.co) platform help — a consumer user-research marketplace that both recruits participants and runs the session: live moderated video conversations, unmoderated test recordings, unlimited async messaging, and surveys across a 50k+ community filtered on 100+ screening criteria. Its edge is price — a flat session rate that includes the participant incentive, where recruit-only marketplaces bill a fee plus incentive on top; bringing your own users is free. UI-only: no public API, webhooks, Zapier, or MCP. Use when recruiting consumer testers on a shoestring, working out whether the free or paid plan is cheaper, participants turning out to be a bad fit, stale participant profile data, no-shows with no reminder or consent workflow, messaging users continuously instead of one-off studies, or choosing between Lightster and Respondent / User Interviews / UserTesting. Do NOT use for the validate-before-building method (use /sales-idea-validation) or B2B recruitment (use /sales-respondent).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
The command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
File Explorer
3 files
Showing SKILL.md
SKILL.md
Source instructions · Read-only preview
name
sales-lightster
description
Lightster (lightster.co) platform help — a consumer user-research marketplace that both recruits participants and runs the session: live moderated video conversations, unmoderated test recordings, unlimited async messaging, and surveys across a 50k+ community filtered on 100+ screening criteria. Its edge is price — a flat session rate that includes the participant incentive, where recruit-only marketplaces bill a fee plus incentive on top; bringing your own users is free. UI-only: no public API, webhooks, Zapier, or MCP. Use when recruiting consumer testers on a shoestring, working out whether the free or paid plan is cheaper, participants turning out to be a bad fit, stale participant profile data, no-shows with no reminder or consent workflow, messaging users continuously instead of one-off studies, or choosing between Lightster and Respondent / User Interviews / UserTesting. Do NOT use for the validate-before-building method (use /sales-idea-validation) or B2B recruitment (use /sales-respondent).
argument-hint
[describe what you need help with in Lightster]
license
MIT
version
1.0.0
tags
["sales","pre-launch","platform"]
Lightster Platform Help
Lightster (lightster.co) is a consumer user-research marketplace with an unusual shape: it both
recruits participants and runs the session, in one tool, at the cheapest per-session price in
the cluster. Researchers are Creators; participants are Lightsters (a 50,000+ community
filterable on 100+ criteria plus custom screening questions). Four methods: Live Conversations
(scheduled moderated video, screen sharing, recording), Unmoderated Testing (reaction videos),
Unlimited Messaging (async text with matched users), and Surveys. Recordings carry AI voice
transcription plus extracted key actions. Web app plus iOS/Android.
Three things to say almost every time:
The session price includes the incentive — that's the whole pricing story. Recruit-only
marketplaces (Respondent, User Interviews) bill a recruit fee plus the incentive on top; Lightster
quotes one flat number with the reward inside it. Compare all-in per session, never fee-vs-fee.
The subscription is a session discount, not an access fee. The free plan's sessions cost several
times the paid plan's, so the fee pays for itself within a couple of sessions — the free tier is the
more expensive choice at real volume. Do that arithmetic before recommending a plan.
Lightster is UI-only.No public API, webhooks, Zapier, or MCP — "pipe transcripts into my
warehouse" is manual export, not REST. Route pipeline needs to /sales-userintuition.
Step 1 — Gather context
If references/learnings.md exists, read it first for accumulated platform knowledge.
Ask only what you can't infer:
What do you want from Lightster?
A) Pick a method — Live Conversation vs Unmoderated Test vs Messaging vs Survey
B) Budget it — free vs paid plan, what a session actually costs all-in
C) Recruit — the 50k community and 100+ criteria vs bringing your own users (free)
D) Fix participant quality — bad-fit participants, stale profiles, no-shows
E) Get data out — export (no API)
F) Choose — Lightster vs Respondent / User Interviews / UserTesting / Great Question
Are your users consumers or B2B professionals? Lightster's community is consumer-weighted. For
hard-to-reach B2B/professional audiences, Respondent is the stronger panel — say so plainly.
Are you testing something built, or still deciding whether to build? The second is an
idea-validation question — flag it in Step 2.
Skip-ahead: if the user's prompt already names the method and the question is specific, go to Step 3.
Step 2 — Route or answer directly
If the user's question is about…
Route to
Comparing research/validator tools across the market, or the validate-before-building method
/sales-idea-validation {question}
Recruiting B2B / professional / hard-to-reach participants
/sales-respondent {question}
An API / webhook / MCP-native research pipeline (Lightster has none)
/sales-userintuition {question}
A card sort / tree test / first-click IA study (Lightster has no IA battery)
/sales-uxtweak or /sales-lyssna{question}
An end-to-end research-ops suite with a repository
/sales-great-question {question}
Running a real behavior demand test (smoke page, waitlist, pre-sale)
/sales-idea-validation or /sales-funnel{question}
Analyzing existing NPS/CSAT/VoC feedback (post-launch, not new sessions)
/sales-customer-feedback {question}
When routing, give the exact command: "This is a {domain} question — run: /sales-respondent {original question}"
Otherwise, answer Lightster-specific questions using Step 3.
Step 3 — Lightster platform reference
Read references/platform-guide.md for the full reference — the method catalog, the
module/automation-surface table, best-effort pricing with the incentive-inclusive session math, the
community's composition and screening surface, the Creator → Criteria → Match → Session data model,
and the no-API data-out playbook. Lightster has no public API, so there is no API-reference file.
Answer using only the relevant section — don't dump the full reference.
Step 4 — Actionable guidance
Never quote a session price without saying the incentive is already inside it. This applies in
every context — a plan comparison, a budget question, a single quoted number, with no competitor
mentioned at all — not just when comparing against another tool. Lightster's session price already
contains the participant reward, so nothing gets added later; say so in the same sentence as the
number. It is the platform's defining commercial fact and the thing users most need to hear.
When another tool is in play, compare all-in vs all-in. Respondent, User Interviews, and
UserTesting quote a recruit fee and add the incentive (often the larger half) on top. Comparing
Lightster's number to a peer's fee understates the peers by a wide margin — the most common
mistake here. Whenever you quote any figure, label it best-effort in the same breath and point to
lightster.co/pricing — these numbers move and a bare figure reads as verified fact.
Do the free-vs-paid arithmetic before naming a plan, and expect it to favour paid. The paid
plan's per-session price is a large multiple cheaper than the free plan's, so the monthly fee
amortizes after roughly two sessions. Say the break-even out loud rather than defaulting to "start
free" — for anyone running more than a couple of sessions the free tier costs more in total. Both
plans' session prices are incentive-inclusive, so state that when you show the math — the
numbers are already all-in and nothing gets added later. The free tier's real use is a single
exploratory session or messaging a few users, and it also caps you at a handful of screening
criteria versus the full 100+.
Say that bringing your own users is free on every plan. If the user already has customers, a
waitlist, or a community, self-recruiting removes the session cost entirely and they still get the
scheduling, recording, transcription, and AI-actions machinery. Lead with this for a founder
testing with existing users — it is the cheapest path in the whole cluster and it's easy to miss.
Treat bad-fit participants as a screener-design problem, not a bad panel. The #1 reported pain
is not knowing whether someone fits until you're in the session, compounded by profile data that
goes stale. Tell users to write behavior-based custom screening questions (what they did
recently, not what they'd claim about themselves), use hidden disqualifiers, and re-verify the
key criterion in the first minute of the session rather than trusting the profile. On the paid
plan, double-screen with free messaging before booking — message the match, confirm fit, then
spend the session. That messaging step is the platform's actual answer to this complaint and is the
single highest-leverage habit to recommend. Tell them that if a participant still turns out not to
fit, end the session early rather than burning thirty minutes collecting unusable data. And
whenever participant quality comes up, add the adjacent warning even if they didn't ask about it:
there are no automated appointment reminders and no pre-interview consent workflow, so they
should over-recruit to absorb no-shows and handle consent and reminders themselves in the
messaging thread — the same thread they're already using to double-screen.
Warn that there are no consent or reminder workflows, and say what to do instead. Reviewers
report no pre-interview consent step and no appointment reminders — so no-shows and consent gaps are
the researcher's job. Tell them to send consent in the messaging thread ahead of time, send their
own reminder the day before and an hour ahead, and over-recruit to absorb no-shows.
Position it as consumer-weighted, and name Respondent for B2B. The community's strength is
everyday consumers; for hard-to-reach professionals by job title/seniority/firmographics, say
plainly that Respondent's B2B panel is the better instrument and route there.
Flag it as UI-only whenever automation or export comes up.No public API, webhooks, Zapier, or
MCP (verify against current docs — it may change). Data-out is manual export of recordings and
transcripts, so a warehouse or dashboard leg means a scheduled human export plus the user's own
ETL — say plainly that it won't be real-time, and note the AI transcription and extracted key
actions are readable in the app but don't come out programmatically. Don't design a REST
pipeline around it; route that requirement to /sales-userintuition (or /sales-great-question if
a repository matters too).
Say the caveat: a session is not demand. Watching people react and hearing why is the
talk-to-users rung of the evidence ladder — a participant saying "I'd pay" is not a purchase.
Keep the why, and take the build-or-not go/no-go to a real behavior test via
/sales-idea-validation.
Distrust the panel-size headline and say which number you're using. The site's own figures
disagree (the homepage says 10,000+, review sites and the vendor profile say 50,000+), so treat
community size as approximate and never let a study plan depend on it — for a narrow demographic,
tell the user to run the criteria filter first and see the actual match count before committing.
If you discover a gotcha or tip not in references/learnings.md, append it there with today's date.
Gotchas
Best-effort from research (2026-07) — pricing, plan gates, community size, and the (absent) API/export surface move; verify at lightster.co.
The session price includes the incentive; the peers' prices don't. Lightster's per-session number has the participant reward baked in. Respondent/User Interviews/UserTesting quote a recruit fee and the incentive you set (for B2B often the larger component) is extra. Any comparison that isn't all-in-vs-all-in is wrong.
The free plan makes each session several times more expensive. Starter's sessions start far above Premium's, so the subscription is a session discount — it pays for itself in ~2 sessions. "Start free" is the wrong default for anyone with real volume.
The free plan caps screening at a few criteria. Starter allows a handful of criteria and messaging with only a small number of users per month; the 100+ criteria, unlimited screening questions, and double-screen messaging are paid-plan features. Narrow audiences are effectively paid-only.
Bringing your own users is free on every plan. Own customers/waitlist/community cost nothing per session and still get scheduling, recording, transcription, and AI actions. Easy to miss and often the cheapest answer.
Participant fit is unknowable until you talk to them, and profiles go stale. The top reported pain. Behavior-based screeners, hidden disqualifiers, double-screening via messaging before booking, and re-verifying in the session's first minute are the mitigations.
No pre-interview consent workflow and no appointment reminders. Reviewers call both out. Send consent through the messaging thread and run your own reminders; over-recruit for no-shows.
The community is consumer-weighted. For B2B/professional targeting by job title, seniority, or firmographics, Respondent is the stronger panel — route rather than force it.
No API, no webhooks, no Zapier, no MCP. UI-only (web + iOS/Android); data-out is manual export. Route live pipelines to /sales-userintuition.
The panel-size claim is internally inconsistent — the homepage says 10,000+ while review sites and the vendor profile say 50,000+. Treat it as approximate; check the real match count with the criteria filter before planning around it.
Parts of the product are still manual, and reviewers say so. Small team, early-stage: expect rough edges, manual steps, and a short learning curve. Managed recruitment (with ~1-week turnaround) is a Professional-tier service, not self-serve speed.
The lightster.co blog is SEO-hijacked with unrelated spam. Its content farm carries off-topic casino articles. Don't cite the blog as documentation — use the pricing and product pages.
Related skills
/sales-respondent — The recruit-only marketplace peer with a genuine B2B/professional edge and a real Partner REST API — route here when the audience is hard-to-reach professionals or the recruitment must be embedded programmatically; note Respondent bills a recruit fee plus incentive where Lightster's session price includes it. Install: npx skills add sales-skills/sales --skill sales-respondent -a claude-code
/sales-great-question — The research-ops suite peer that recruits, runs any method, and stores a repository (REST API + webhooks + MCP), at a per-seat price — route here when the user wants an end-to-end research operation rather than the cheapest session. Install: npx skills add sales-skills/sales --skill sales-great-question -a claude-code
/sales-userbrain — The cheap think-aloud video peer: only task-based unmoderated recordings on a large tester panel — route here when they want to watch people use a built thing and don't need live conversations or messaging. Install: npx skills add sales-skills/sales --skill sales-userbrain -a claude-code
/sales-uxtweak — The usability + IA suite with card sorting, tree testing, and first-click tests, which Lightster has none of — route here for information-architecture questions. Install: npx skills add sales-skills/sales --skill sales-uxtweak -a claude-code
/sales-lyssna — The ex-UsabilityHub usability + IA suite with a large panel and a usable free plan — the peer pick when the job is testing a design rather than talking to users. Install: npx skills add sales-skills/sales --skill sales-lyssna -a claude-code
/sales-userintuition — The API/webhook/MCP-native real-human interview peer — the pick when a programmatic research pipeline is a requirement Lightster can't meet. Install: npx skills add sales-skills/sales --skill sales-userintuition -a claude-code
/sales-idea-validation — The tool-agnostic validate-before-building method plus the full research/validator tool landscape (use this to actually decide build-or-not; a session saying "I'd pay" is not demand). Install: npx skills add sales-skills/sales --skill sales-idea-validation -a claude-code
/sales-customer-feedback — Real-customer NPS/CSAT/VoC strategy across tools (use to analyze existing feedback; Lightster proactively runs new sessions). Install: npx skills add sales-skills/sales --skill sales-customer-feedback -a claude-code
/sales-funnel — Build the smoke-test / fake-door landing page that turns a research hypothesis into a real demand test. Install: npx skills add sales-skills/sales --skill sales-funnel -a claude-code
/sales-do — Not sure which skill to use? The router matches any sales objective to the right skill. Install: npx skills add sales-skills/sales --skill sales-do -a claude-code
Examples
Example 1: "I need to talk to 10 real users and I'm nearly broke"
User says: "Solo founder, consumer app idea. I want to interview 10 people in my target demo. UserTesting quoted me five figures. Is Lightster actually cheap or is that a bait price?"
Skill does: Confirms it's genuinely cheap and explains why: the quoted session price includes the participant incentive, where the peers bill a recruit fee and add the reward on top — so the gap is real, not a bait. Then does the plan arithmetic out loud: the paid plan's sessions cost a fraction of the free plan's, so for 10 sessions the subscription pays for itself almost immediately and "start free" would cost more. Asks whether the founder has any existing users or waitlist — if so, bringing your own users is free on every plan, which beats every option. Flags all figures as best-effort against lightster.co/pricing, and closes with the caveat that 10 people saying "I'd pay" is not a purchase; the go/no-go belongs to a real behavior test via /sales-idea-validation.
Result: The founder runs 10 sessions on the paid plan for roughly what one UserTesting session would have cost, and knows what the sessions do and don't prove.
Example 2: "Can I pipe Lightster transcripts into our warehouse?" (developer/automation)
User says: "I want every completed session's transcript to land in BigQuery automatically so we can tag themes across studies."
Skill does: States plainly that Lightster is UI-only — no public API, webhooks, Zapier, or MCP — there is no automated path; data-out is manual export of recordings and transcripts, so a warehouse leg means someone exporting and an ETL job over the drop, and it won't be real-time. Notes the AI transcription and extracted key actions are readable in the UI but don't come out programmatically. Says this is the platform's central trade-off — it buys the cheapest sessions in the cluster by not being an infrastructure product — and routes a hard programmatic requirement to /sales-userintuition (REST API + HMAC webhooks + MCP), or /sales-great-question if a repository matters, naming the price difference as the trade.
Result: The user stops hunting for an API that doesn't exist and either accepts manual export or switches tools deliberately.
Example 3: "Half my participants weren't who they said they were"
User says: "Ran four sessions on Lightster. Two were clearly not in my target market — I only found out five minutes in. Is the panel just bad?"
Skill does: Reframes it as a screener-design problem rather than a bad panel — this is the platform's #1 reported pain, and it's compounded by participant profile data going stale. Prescribes the fixes: behavior-based custom screening questions (what they actually did recently, not self-description), hidden disqualifiers, and the highest-leverage habit — double-screen via messaging before booking (a paid-plan feature: message the match, confirm fit, then spend the session), plus re-verifying the key criterion in the session's first minute. Notes the free plan's few-criteria cap makes this worse and is part of why they're seeing it. Adds that Lightster has no reminder or consent workflow, so those are their job too. If the target is B2B professionals, says plainly the community is consumer-weighted and routes to /sales-respondent.
Result: The user rewrites the screener behaviorally and double-screens by message; fit problems largely stop.
Troubleshooting
"Should I just start on the free plan?"
Symptom: The user assumes free is the cheap entry point and paid is an upsell.
Cause: Lightster's subscription is a session discount, not an access fee — the free plan's sessions start at several times the paid plan's price, and free also caps screening at a few criteria and messaging to a small number of users per month.
Solution: Multiply sessions × per-session price under each plan and compare totals including the subscription. The fee amortizes in roughly two sessions, so free is only right for a single exploratory session or a bit of messaging; past that it costs more. Also check whether they can bring their own users — that's free on every plan and beats both. Confirm current figures at lightster.co/pricing and present them as best-effort.
"I can't tell if a participant is right until I'm in the call"
Symptom: Sessions burned on people who don't match the target; profile data looks out of date.
Cause: The top reported pain — self-reported profiles drift, and criteria filtering alone doesn't prove fit.
Solution: Write behavior-based custom screening questions (recent actions, not self-description), add hidden disqualifiers, and — the real fix — double-screen with messaging before booking on the paid plan: message the match, confirm fit in the thread, then spend the session. Re-verify the key criterion in the first minute. Over-recruit to absorb no-shows, since there are no automated reminders. If the audience is B2B/professional rather than consumer, the panel genuinely is the wrong instrument — route to /sales-respondent.
"How do I get sessions and transcripts out automatically?"
Symptom: The user wants an integration, a Zap, or an API key and can't find one.
Cause: Lightster is UI-only — no public API, no webhooks, no Zapier, no MCP, no developer portal.
Solution: Data-out is manual: download recordings, copy transcripts, export results. Build any warehouse or dashboard leg on a scheduled human export plus your own ETL, and don't promise real-time. If a live programmatic pipeline is a hard requirement, route to /sales-userintuition (REST API + HMAC-signed webhooks + MCP server) or /sales-great-question (API + webhooks + MCP + repository), and name the cost difference as the deliberate trade-off.
