8 Comment Prompts to Pin on Your Short Videos

What you paste, what you get

You paste: one paragraph about a short vertical video you plan to post (topic, what it shows, who watches, tone, language, and where the question will go).

You get: exactly 8 comment prompts, each with its type, what the viewer is asked to write, the part of the short that makes the question natural, a counted character length and a bait check. After the list come a validation table, four review lenses, applied in turn by one model, and a ranked top 3. It plans; it does not watch videos or read comments. Validation and review lenses are self-checked by the same model; spot-check the numbers.

$5 one-time download. Sign in, pay once and download the SKILL.md, then run it in your own AI agent. Built for Claude Code; other agents that read SKILL.md files (such as Codex CLI or OpenClaw) should also work. No network access or extra tools needed. No subscription from us: you use your own agent and its plan.

GlowHalo is not affiliated with or endorsed by Anthropic, OpenAI or OpenClaw; the names are used only to say which agents the file is written for.

Sample output

Sample input (an invented example)
I am posting a 25-second vertical video on my small channel about home plant care. It shows three ways to save an over-watered houseplant: stop watering for a week, repot into dry soil, and trim the soft brown roots. The last 5 seconds show the plant a month later with new leaves. Viewers are beginners who own one or two houseplants. Tone: calm and friendly. Language: English. I plan to put the question in a pinned comment. Avoid the words "guys" and "hack".

Real, unedited output from running this Skill on a sample problem:

Planning only: this cannot see comments, analytics or how any platform shows or ranks videos - nothing here predicts comments, reach or engagement.

Short: three ways to save an over-watered houseplant (25 s) · main hook: the three fixes and the plant a month later · English · pinned comment · 8 prompts · SFW.

## Essentials

- Topic: home plant care, saving an over-watered houseplant `(given)`
- What it shows: three fixes (stop watering for a week, repot into dry soil, trim the soft brown roots), then the plant a month later with new leaves in the last 5 seconds `(given)`
- Length: 25-second vertical video, small channel `(given)`
- Audience: beginners who own one or two houseplants `(given)`
- Tone: calm and friendly `(given)`
- Language: English `(given)`
- Placement: pinned comment `(given)`
- Avoid: "guys", "hack" `(given)`
- Emoji: none (not requested) `(assumed)`

## Hooks map

- **H1** The choice between three fixes for one problem.
- **H2** Stop watering for a week. This is a step that asks for patience.
- **H3** Repot into dry soil.
- **H4** Trim the soft brown roots. Cutting a sick plant can feel risky.
- **H5** The plant a month later with new leaves, which shows it can recover.
- **H6** Over-watering is the mistake behind the whole short.

Anchors:
- Audience: beginners with one or two houseplants.
- Tone: calm and friendly. No pressure, no teasing.
- Prompt language: English.
- Placement: pinned comment.
- Avoid list (exact words): "guys", "hack".
- Working limits: a prompt is 100 characters or fewer including spaces. For on-screen use, 60 or fewer is a planning estimate for a small phone overlay. Real limits vary by app and screen.

## Prompt set

| # | Prompt | Type | Asks for | Hangs on | Chars | Placement fit | Flags |
|---|---|---|---|---|---|---|---|
| 01 | Which fix would you try first: a week without water, repotting, or trimming roots? | T this-or-that | a pick (one of three) | H1 | 82 | pinned | 3a CORROBORATED (applied) |
| 02 | Have you ever lost a plant to too much water? A yes or no is fine. | E personal experience | yes or no | H6 | 66 | pinned | 4a DISPUTED, 3b NO CHANGE |
| 03 | Trimming roots on a sick plant: does it sound scary or sensible to you? | O quick opinion | one word (scary / sensible) | H4 | 71 | pinned | none |
| 04* | Finish the line: before this video, I thought soft brown roots meant ___. | B fill-in-the-blank | a few words | H4 | 73 | pinned | Phase 3 CLEAR fix |
| 05* | Your turn: what is one tip, old or just learned, to avoid over-watering? | S tip swap | one short sentence | H6, H2 | 72 | pinned | 1a CORROBORATED (applied) |
| 06 | Repotting into dry soil: had you heard of that fix before today? | Q topic question | yes or no | H3 | 64 | pinned | none |
| 07* | From 1 to 5, how sure are you that a soggy plant can be saved? | R rate | a number 1-5 | H5 | 62 | pinned | 3a CORROBORATED (applied) |
| 08 | Prediction: could you leave the watering can alone for a full week? Yes or no. | P prediction | yes or no | H2 | 78 | pinned | 3b NO CHANGE |

`*` = changed after the first draft. Row 04 was changed in Phase 3. Rows 05 and 07 were changed by the review lenses.

## Validation

**1. Count and uniqueness: PASS.**
- There are 8 rows.
- The first words are all different: Which, Have, Trimming, Finish, Your, Repotting, From, Prediction.
- Each row asks for something different.
- The set uses 8 types, and no type appears on more than 1 row.

**2. Length [LEN]: PASS.** All prompts are 100 characters or fewer. Counts for the final set:
- 01: Which(5) fix(3) would(5) you(3) try(3) first:(6) a(1) week(4) without(7) water,(6) repotting,(10) or(2) trimming(8) roots?(6) = 69 + 13 spaces = 82
- 02: Have(4) you(3) ever(4) lost(4) a(1) plant(5) to(2) too(3) much(4) water?(6) A(1) yes(3) or(2) no(2) is(2) fine.(5) = 51 + 15 spaces = 66
- 03: Trimming(8) roots(5) on(2) a(1) sick(4) plant:(6) does(4) it(2) sound(5) scary(5) or(2) sensible(8) to(2) you?(4) = 58 + 13 spaces = 71
- 04: Finish(6) the(3) line:(5) before(6) this(4) video,(6) I(1) thought(7) soft(4) brown(5) roots(5) meant(5) ___.(4) = 61 + 12 spaces = 73
- 05: Your(4) turn:(5) what(4) is(2) one(3) tip,(4) old(3) or(2) just(4) learned,(8) to(2) avoid(5) over-watering?(14) = 60 + 12 spaces = 72
- 06: Repotting(9) into(4) dry(3) soil:(5) had(3) you(3) heard(5) of(2) that(4) fix(3) before(6) today?(6) = 53 + 11 spaces = 64
- 07: From(4) 1(1) to(2) 5,(2) how(3) sure(4) are(3) you(3) that(4) a(1) soggy(5) plant(5) can(3) be(2) saved?(6) = 48 + 14 spaces = 62
- 08: Prediction:(11) could(5) you(3) leave(5) the(3) watering(8) can(3) alone(5) for(3) a(1) full(4) week?(5) Yes(3) or(2) no.(3) = 64 + 14 spaces = 78

**3. Fit [FIT]: n/a for this short.**
- The placement is pinned, so the on-screen limit does not apply.
- For reference, all 8 prompts are over 60 characters. None should be moved on screen without shortening.

**4. Hangs on a hook [HOOK]: PASS.** All 8 rows are FULL:
- 01 → H1
- 02 → H6
- 03 → H4
- 04 → H4
- 05 → H6, H2
- 06 → H3
- 07 → H5
- 08 → H2

**5. Easy to answer [EASY]: PASS.**
- Every row can be answered in one line: a pick, yes or no, one word, a number, or one short sentence.
- No row asks for personal data, a location, a link or a long story.

**6. Bait check [BAIT]: PASS.**
- No row includes gating, deals, rewards, urgency, guilt or like/follow demands.

**7. Stranger test [CLEAR]: PASS after one fix.**
- Row 04's draft said "before this," with an undefined "this" (WARN). It was fixed to "before this video,".

**8. Avoid list, tone and SFW: PASS.**
- No row contains "guys" or "hack", or a respelling of either.
- The tone is calm throughout.
- No row blames a group.

**Phase 3 change list:**
- Row 04, CLEAR fix:
  - Before: "Finish the line: before this, I thought soft brown roots meant ___." (67)
  - After: "Finish the line: before this video, I thought soft brown roots meant ___." (73)

**Counts line:**
- 8 rows.
- 8 types used [T, E, O, B, S, Q, R, P].
- 0 prompts fit on screen. All are over 60, which is fine because the placement is pinned.
- Longest prompt: 01 at 82 characters.

### Validation table

| # | Chars | Fit | Hook | Easy | Bait | Clear | Avoid/SFW |
|---|---|---|---|---|---|---|---|
| 01 | 82 | n/a (pinned) | H1 FULL - PASS | PASS | PASS | PASS | PASS |
| 02 | 66 | n/a (pinned) | H6 FULL - PASS | PASS | PASS | PASS | PASS |
| 03 | 71 | n/a (pinned) | H4 FULL - PASS | PASS | PASS | PASS | PASS |
| 04* | 73 | n/a (pinned) | H4 FULL - PASS | PASS | PASS | PASS (fixed) | PASS |
| 05* | 72 | n/a (pinned) | H6, H2 FULL - PASS | PASS | PASS | PASS | PASS |
| 06 | 64 | n/a (pinned) | H3 FULL - PASS | PASS | PASS | PASS | PASS |
| 07* | 62 | n/a (pinned) | H5 FULL - PASS | PASS | PASS | PASS | PASS |
| 08 | 78 | n/a (pinned) | H2 FULL - PASS | PASS | PASS | PASS | PASS |

## Review lenses

### Findings per lens

**Lens 1 - Scrolling Viewer**
- **1a [05]** Draft row 05 read: "Your turn: what is one small habit that keeps you from over-watering?"
  - A beginner with one or two plants may not have a habit yet, so the question feels like homework.
  - Fix: let them share a tip they just learned from the short.

**Lens 2 - Honesty Auditor**
- No findings. No bait, deals, hidden gating or invented rewards. No row implies content the short does not contain.

**Lens 3 - Wording Checker**
- **3a [01, 07]** Row 01 reads "Which fix would you try first: a week without water, repotting, or trimming roots?" and draft row 07 read "Rank the three fixes, easiest to hardest: no water, repot, trim roots."
  - Both rows list the same three fixes and both ask for an order or a pick, so they look alike.
  - Fix: rewrite row 07 to rate the month-later result (H5) instead.
- **3b [02, 08]** Row 02 ends "A yes or no is fine." and row 08 ends "Yes or no."
  - The endings are similar, but the questions ask different things (a past experience versus a prediction).
  - Flag only. No change proposed.

**Lens 4 - Tone & Safety Scout**
- **4a [02]** Row 02 reads "Have you ever lost a plant to too much water? A yes or no is fine."
  - "Lost" may feel like admitting failure to a beginner.
  - Proposed: "Has too much water ever made one of your plants droop? A yes or no is fine."

### Cross-check

- `[1a, 05] Lens 2: no challenge / Lens 3: no challenge / Lens 4: corroborate - "asking beginners for a habit they may not have puts them on the spot; 'old or just learned' keeps the calm tone"`
- `[3a, 01, 07] Lens 1: corroborate - "two rows listing the same three fixes read as the same question twice in a two-second glance" / Lens 2: no challenge / Lens 4: no challenge`
- `[3b, 02, 08] Lens 1: no challenge / Lens 2: no challenge / Lens 4: no challenge`
- `[4a, 02] Lens 1: challenge - "'lost' is plain and the line already lowers pressure with 'A yes or no is fine.'; 'droop' asks a vaguer thing that matches H6 less directly" / Lens 2: no challenge / Lens 3: no challenge`

### Verdict

- **1a CORROBORATED (applied).** Raised by Lens 1 and corroborated by Lens 4 in the cross-check for 1a. Row 05 was rewritten.
- **3a CORROBORATED (applied).** Raised by Lens 3 and corroborated by Lens 1 in the cross-check for 3a. Row 07 was rewritten; row 01 was kept as is.
- **3b NO CHANGE.** No lens proposes a change. It stays as a flag on rows 02 and 08.
- **4a DISPUTED.** Row 02 is left as written.
  - Lens 4: "lost" may sting a beginner, so it could be softened to "Has too much water ever made one of your plants droop? A yes or no is fine."
  - Lens 1: "lost" is plain, the yes/no line already keeps it low-pressure, and "droop" is vaguer.
  - The user decides.

Of the 4 findings, 2 were applied [1a, 3a], 1 is disputed [4a] and 1 needs no change [3b].

### Changes applied

**Row 05 (1a):**
- Prompt before: "Your turn: what is one small habit that keeps you from over-watering?" (69)
- Prompt after: "Your turn: what is one tip, old or just learned, to avoid over-watering?" (72)
- Asks for "one short sentence" and Hangs on "H6, H2" are unchanged.
- Phase 3 re-run on 05: LEN 72 PASS, HOOK H6, H2 FULL PASS, EASY PASS, BAIT PASS, CLEAR PASS, Avoid/SFW PASS. The first word "Your" is still unique.

**Row 07 (3a):**
- Prompt before: "Rank the three fixes, easiest to hardest: no water, repot, trim roots." (70)
- Prompt after: "From 1 to 5, how sure are you that a soggy plant can be saved?" (62)
- Asks for before: "an order of the three fixes". After: "a number 1-5".
- Hangs on before: "H1". After: "H5".
- Type "R" is unchanged.
- Phase 3 re-run on 07: LEN 62 PASS, HOOK H5 FULL PASS, EASY PASS, BAIT PASS, CLEAR PASS, Avoid/SFW PASS. The first word changed from "Rank" to "From", which is still unique. The type rule still holds.

## Top 3 (ranked)

1. **Row 01**, T this-or-that: "Which fix would you try first: a week without water, repotting, or trimming roots?"
   - Why: it uses the whole short (all three fixes from H1), and the answer is a single pick a beginner can give in one word.
   - Attached: 3a CORROBORATED (applied).
2. **Row 06**, Q topic question: "Repotting into dry soil: had you heard of that fix before today?"
   - Why: it names the fix up front, so a mid-scroll viewer understands it. The answer is yes or no, and it suits beginners.
   - Attached: clean.
3. **Row 03**, O quick opinion: "Trimming roots on a sick plant: does it sound scary or sensible to you?"
   - Why: it speaks to the one step beginners may hesitate over (H4), and it offers two one-word answers in a calm tone.
   - Attached: clean.

## Notes

**Assumptions to confirm:**
- No emoji was used `(assumed)`.
- Row 01 assumes the short presents the three fixes as separate options, which is how the paragraph describes them ("three ways").
- Decide on row 02 (4a): keep "lost a plant", or switch to the softer "droop" wording.

**How to use:**
- Pick one prompt per short and put it in the pinned comment.
- If you later want a prompt on screen, shorten it to about 60 characters first. All 8 are currently longer.
- Change one thing at a time and compare your own numbers over a fair period.

**Honest limit:** this cannot see comments or analytics, and the character limits are planning estimates, so check the real comment box.

**Want more?** I can write 8 more prompts of one type, prompts in another language, or check prompts you already use.

Use cases

• Choosing one honest question to pin under a short instead of "comment YES if you agree"
• Checking a prompt you already use for deals, gating and pressure
• Turning a 25-second how-to into questions beginners can answer in one line
• Keeping a prompt tied to what the short really shows, so it does not imply more

FAQ

Will it get me more comments or reach? No promise either way. It cannot see comments, analytics or how any platform shows videos. It checks that each prompt is easy to answer, tied to your short and free of bait.

Does it offer giveaways or "comment to get" tricks? No. It flags those as bait and rewrites them.

Can it watch my video? No. It trusts your paragraph; if the finished video differs, re-check the prompts.

Which languages? Any, with English shown in the sample. Character limits are planning estimates, and wide scripts get an assumed lower limit.

Is it SFW only? Yes, and medical, legal, financial or political topics are kept light and descriptive.

Not affiliated with or endorsed by YouTube, Google, TikTok or Instagram; platform names are used only to describe what the plan is for.

How it works (in detail)

8 Comment Prompts to Pin on Your Short Videos turns one paragraph about a short vertical video into 8 comment prompts and checks each one before you post. It plans and audits; it does not read comments or predict results.

Phase 1 - Intake and hooks map. Reads what the short shows, who watches, tone, language and placement. Asks at most 3 questions, only if an essential is missing. Lists the moments in your short that give viewers something to react to.

Phase 2 - Draft 8 prompts. Eight prompts of at least five types (this-or-that, personal experience, quick opinion, fill-in-the-blank, tip swap, topic question, rate or rank, prediction), each 100 characters or fewer, each tied to a moment in the short.

Phase 3 - Validation. Real character counts, on-screen fit, hook match, easy-to-answer check, bait check, stranger test, and your avoid list, summed up in a validation table.

Phase 4 - Review lenses. One model applies four lenses in turn: Scrolling Viewer, Honesty Auditor, Wording Checker, Tone & Safety Scout. The other lenses challenge or corroborate each finding, and a verdict tags each CONFIRMED, CORROBORATED, DISPUTED or NO CHANGE. Fixes are applied and listed.

Phase 5 - Ranked top 3. Three prompts of different types, ranked, with reasons drawn from the words only. No promises about comments, reach or engagement.