uplika/uplika-plugin / skills/ig-reel
ig-reel
VerifiedWrite an Instagram Reel from one idea (scored hooks, spoken script, on-screen text, timed beat sheet), then publish or schedule it after approval.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.0 KB |
| scripts/beats.pyexecutable | script | 8.0 KB |
| scripts/hooks.json | config | 17.5 KB |
| scripts/hookscore.pyexecutable | script | 12.5 KB |
scripts/beats.py
View on GitHub#!/usr/bin/env python3
"""
beats.py - turn a Reel script into a timed beat sheet before you record it.
Estimates how long each line takes to say, stacks them into timecodes, and
flags the four things that kill a Reel in the edit: a hook that runs past the
three-second mark, a beat long enough for the viewer to leave, a run of lines
with nothing concrete in them, and a total length that does not match what you
said you were making.
The timings are an estimate from word count at a words-per-minute rate. They
are close enough to plan an edit and not a substitute for recording it. Set
your own rate with --wpm once you have timed yourself reading a script out
loud: most people land between 150 and 200, and the default here is 165.
Usage
python3 beats.py script.txt
python3 beats.py script.txt --target 30
python3 beats.py script.txt --wpm 185 --target 45
pbpaste | python3 beats.py -
python3 beats.py script.txt --json
"""
import argparse
import json
import re
import sys
WORD_RE = re.compile(r"[A-Za-z0-9$%'’-]+")
SENT_RE = re.compile(r"[^.!?]+[.!?]*")
CONCRETE_RE = re.compile(r"\$\s?\d|\b\d[\d,.]*\b|(?<!^)\b[A-Z][a-z]{2,}\b", re.MULTILINE)
STOPWORDS = {
"the", "a", "an", "and", "or", "but", "if", "of", "to", "in", "on", "for",
"with", "that", "this", "it", "is", "are", "was", "were", "be", "been",
"you", "your", "i", "my", "me", "we", "our", "they", "them", "he", "she",
"so", "just", "not", "no", "do", "did", "does", "have", "has", "had",
"will", "can", "at", "as", "by", "from", "out", "up", "off", "one", "all",
}
HOOK_WINDOW = 3.0 # seconds. Past this, the thumb has already decided.
MAX_BEAT = 4.0 # seconds on one idea with no change on screen.
ABSTRACT_RUN = 3 # beats in a row with nothing checkable in them.
def words(text):
# "$18,000" is one word when it is spoken, so it is one word here too.
return WORD_RE.findall(re.sub(r"(?<=\d),(?=\d)", "", text))
def pretty(token):
"""Put the thousands separator back for display."""
return re.sub(r"(\d)(?=(\d{3})+$)", r"\1,", token)
def tc(seconds):
m, s = divmod(seconds, 60)
return f"{int(m)}:{s:04.1f}"
def split_beats(raw, wps):
"""One line is one beat, unless a line is too long to be one."""
beats = []
for line in [l.strip() for l in raw.splitlines()]:
if not line:
continue
if len(words(line)) / wps <= MAX_BEAT * 1.5:
beats.append(line)
continue
# Long paragraph: break it at sentence ends so the timings mean
# something, and let the report say it was split.
parts = [p.strip() for p in SENT_RE.findall(line) if p.strip()]
buf = ""
for part in parts:
candidate = (buf + " " + part).strip()
if buf and len(words(candidate)) / wps > MAX_BEAT:
beats.append(buf)
buf = part
else:
buf = candidate
if buf:
beats.append(buf)
return beats
def analyse(raw, wpm=165, target=None):
wps = wpm / 60.0
beats = split_beats(raw, wps)
if not beats:
return None
rows, clock = [], 0.0
for i, text in enumerate(beats):
n = len(words(text))
dur = n / wps
rows.append({
"n": i + 1,
"start": round(clock, 2),
"dur": round(dur, 2),
"words": n,
"text": text,
"concrete": len(CONCRETE_RE.findall(text)),
"label": "",
"flags": [],
})
clock += dur
total = clock
# Label the structural positions a Reel is actually built around.
for r in rows:
if r["n"] == 1 or r["start"] + r["dur"] <= HOOK_WINDOW:
r["label"] = "HOOK"
rows[-1]["label"] = "CTA" if rows[-1]["label"] != "HOOK" else "HOOK/CTA"
half = total / 2
for r in rows:
if not r["label"] and r["start"] <= half < r["start"] + r["dur"]:
r["label"] = "MID"
notes = []
if rows[0]["dur"] > HOOK_WINDOW:
rows[0]["flags"].append(f"hook runs {rows[0]['dur']:.1f}s, past the {HOOK_WINDOW:.0f}s mark")
notes.append(f"Beat 1 takes {rows[0]['dur']:.1f}s to say. Cut it to "
f"{int(HOOK_WINDOW * wps)} words or fewer, or the hook lands after "
"the decision has been made.")
if rows[0]["concrete"] == 0:
notes.append("Beat 1 has no number and no name in it. Hooks without something "
"checkable are the ones that get scrolled.")
for r in rows:
if r["dur"] > MAX_BEAT:
r["flags"].append(f"{r['dur']:.1f}s on one beat")
long_beats = [r["n"] for r in rows if r["dur"] > MAX_BEAT]
if long_beats:
notes.append(f"Beat(s) {', '.join(map(str, long_beats))} run past {MAX_BEAT:.0f}s. "
"Either split the line or change what is on screen inside it. "
"A static frame is where people leave.")
run, start = 0, None
for r in rows:
if r["concrete"] == 0:
run += 1
start = start if start is not None else r["n"]
if run == ABSTRACT_RUN:
notes.append(f"Beats {start}-{r['n']} have nothing concrete in them. "
"Put a number, a name or a price in one of them.")
else:
run, start = 0, None
# Does the last line hand you back to the first one?
first = {w.lower() for w in words(rows[0]["text"]) if w.lower() not in STOPWORDS}
last = {w.lower() for w in words(rows[-1]["text"]) if w.lower() not in STOPWORDS}
loop = sorted(pretty(w) for w in first & last)
if loop:
notes.append(f"Loops: the last beat repeats \"{', '.join(loop[:3])}\" from the hook. "
"Second watches are free reach.")
else:
notes.append("No loop. The last beat shares no word with the hook, so the video "
"ends flat. Echoing one word from beat 1 is the cheapest replay you get.")
if target:
delta = total - target
if abs(delta) <= target * 0.1:
notes.append(f"Length is on target ({total:.1f}s against {target}s).")
elif delta > 0:
notes.append(f"{delta:.1f}s over target. Cut about {int(delta * wps)} words.")
else:
notes.append(f"{-delta:.1f}s under target. Either add {int(-delta * wps)} words "
"or shoot it short. Short is usually right.")
return {
"wpm": wpm, "target": target,
"total_seconds": round(total, 2),
"total_words": sum(r["words"] for r in rows),
"beats": rows,
"notes": notes,
}
def render(a, out=sys.stdout):
head = (f"BEAT SHEET · {a['total_words']} words · ~{a['total_seconds']:.1f}s "
f"at {a['wpm']} wpm" + (f" · target {a['target']}s" if a["target"] else ""))
print("\n" + head, file=out)
print("=" * max(len(head), 72), file=out)
for r in a["beats"]:
label = f"{r['label']:<8}" if r["label"] else " " * 8
print(f" {tc(r['start'])} {r['dur']:4.1f}s {label}{r['text']}", file=out)
for f in r["flags"]:
print(f" {'':>6} {'':>5} {'':<8}^ {f}", file=out)
print("-" * max(len(head), 72), file=out)
for n in a["notes"]:
print(f" - {n}", file=out)
print("", file=out)
def main():
ap = argparse.ArgumentParser(description="Time a Reel script into a beat sheet.")
ap.add_argument("input", nargs="?", default="-", help="script file, or - for stdin")
ap.add_argument("--wpm", type=float, default=165, help="speaking rate (default 165)")
ap.add_argument("--target", type=float, help="target length in seconds")
ap.add_argument("--json", action="store_true")
args = ap.parse_args()
raw = sys.stdin.read() if args.input == "-" else open(args.input, encoding="utf-8").read()
a = analyse(raw, wpm=args.wpm, target=args.target)
if not a:
print("empty script", file=sys.stderr)
sys.exit(2)
if args.json:
print(json.dumps(a, indent=2, ensure_ascii=False))
else:
render(a)
sys.exit(0)
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |