agentleFS
lorena-bordonaba-paulorena-bordonaba-pau/youtube-agent/.claude/skills/competitor-scan/SKILL.md

competitor-scan

Skill2 starsChanged 49 days agoMITView on GitHub

The weekly scan of competitor, inspiration and neighbourhood channels — what is performing right now, whether it already exists in your language, and the 3 widest gaps. Run by /radar. Triggered by "run the weekly scan", "what are my competitors doing this week", "scan my lists". In Spanish, "pasa el escaneo semanal", "qué están haciendo mis competidores esta semana", "escanea mis listas".

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Not the competitor board (the web page with every video of the last 6 months: that is competitor-board) and not a full ideation batch (that …

    line 13
  2. Distinctive:

    Read CLAUDE.md if you have not already done so this session.

    line 18
  3. Very specific:

    If fewer than 10 outliers come back, drop to --min-ratio 1.5 before concluding there is nothing.

    line 32
  4. Distinctive:

    If you have a memory about how to read outliers, apply it here.

    line 57
  5. Distinctive:

    Before an outlier goes into the table, check its likes per 1k views against the channel's usual.

    line 74
  6. Very specific:

    Read memory/rules.md and memory/creator_profile.md.

    line 80
  7. Distinctive:

    Title reminder: do not copy the original's formula.

    line 86
  8. Very specific:

    If an outlier changes what was known about the niche, propose updating memory/.

    line 106
  9. Very specific:

    If there is a competitor board (memory/competitor_board.md), sync it with the competitor-board skill and end with its link.

    line 108

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Weekly competitor scan

WHEN

The weekly review of the channel lists, usually through /radar. Always started by the creator.

Not the competitor board (the web page with every video of the last 6 months: that is competitor-board) and not a full ideation batch (that is competitor-ideation). This scan finds what moved recently and where the gap is; it ends with three proposals, not eight.

Read CLAUDE.md if you have not already done so this session.

EXECUTION ORDER

1. Scan

python3 tools/yt_outliers_channels.py --min-ratio 2.0 --sample 25 --max-days 45

--max-days 45 is not optional. Without it, ranking by ratio buries what is recent under outliers from 6-11 months ago, which are no longer opportunities.

If fewer than 10 outliers come back, drop to --min-ratio 1.5 before concluding there is nothing.

2. Add what was saved by hand

python3 tools/yt_outliers_playlist.py --days 14

3. Split by list — each one is read differently

  • competitors (same language, same niche): what works here is direct competition. If a competitor lands an outlier, the gap is closing.
  • inspiration (often another language, an arbitrage source): this is usually the list that matters. For each outlier, the question is: does it already exist in your language?
  • neighbourhood (channels YouTube associates with yours): tells you how the algorithm is classifying you.

4. Adjust for market size before quoting figures

Views in a larger-market language are not reachable as-is in a smaller one. Measure your own factor over time and write it into memory/; until you have one, say you are quoting the raw figure and that it overstates the expectation.

If you have a memory about how to read outliers, apply it here.

5. Check the gap in your language

For the 3-5 strongest foreign-language outliers:

python3 tools/kw_research.py --kw "topic in your language; variant"
python3 tools/yt_search.py --query "topic in your language"

yt_search costs 100 quota units per query — use it only on the finalists. And remember it does not filter by language: if the top results are in another language, that does not prove the gap in yours is closed.

6. Leave out bought views

Before an outlier goes into the table, check its likes per 1k views against the channel's usual. Views that arrive without likes are ad reach: drop the video and say so. Affiliate links prove nothing, almost every channel carries them.

7. Filter against the channel's rules

Read memory/rules.md and memory/creator_profile.md. Discard what does not fit and say why.

Format reminder: apply the restrictions in memory/rules.md, if there are any. Never propose a format the channel has discarded.

Title reminder: do not copy the original's formula. Copy the TOPIC and rewrite it with the house formula, whatever your own data says that is. And a high ratio on a huge channel does not mean the same thing as on a small one.

SOURCES

  • Outlier ratios, views and likes: youtube_api (views, likes) and derived (the ratio against the channel's median).
  • kw_research.py scores: heuristic, a proxy — never a search volume.
  • score_titles.py scores: heuristic, never a CTR prediction.
  • The market-size factor: config if it is in memory/, otherwise say there is none and that the raw figure overstates the expectation.

OUTPUT

  1. Opportunity table: outlier, channel, ratio, views, and whether it already exists in your language.
  2. The 3 with the widest gap, each with a proposed title already in the channel's formula, scored with python3 tools/score_titles.py.
  3. One recommendation and the reason for it.
  4. If an outlier changes what was known about the niche, propose updating memory/.
  5. If there is a competitor board (memory/competitor_board.md), sync it with the competitor-board skill and end with its link. If there is none, offer to create one in a single line.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.