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

competitor-board

Skill2 starsChanged 48 days agoMITView on GitHub

Create, sync and read the competitor board, a private web page (an Artifact) listing every long-form video your competitors published in the last 6 months, with thumbnail, date and outlier multiplier, filterable by channel, and with likely ad-bought views flagged. Triggered by "make me a competitor board", "sync the competitor board", "I added competitors to the board", "open the competitor board". In Spanish, "hazme el radar de competidores", "sincroniza el radar de competidores", "he añadido competidores", "abre el radar".

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Read memory/competitor_board.md.

    line 24
  2. Very specific:

    Write memory/competitor_board.md with the URL and the date, and add its line to memory/MEMORY.md.

    line 46
  3. Very specific:

    Read the rows before writing and pin if_version: the user may have edited them.

    line 87

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Competitor board

WHEN

  • The user wants a visual board of what their competitors publish, or asks to create one.
  • "Sync the board", or they say they added or removed @handles on the page.
  • At the end of /radar and competitor-ideation, when a board exists: refresh it so it matches what was just analysed.

The page cannot call the YouTube API on its own (the artifact sandbox blocks it). Handles typed on the page are stored in the page's database and show as pending until this skill runs a sync.

EXECUTION ORDER

0. Is there a board already?

Read memory/competitor_board.md. If it holds a URL, go to Sync. If not, go to Create. Never create a second board when one exists: publishing without its url makes a new page and orphans the list the user built.

Create (first time only)

  1. Build the data from the configured competitors, or from the handles the user gives: python3 tools/competitor_board.py --list competitors python3 tools/competitor_board.py --handles @one @two It adds your own channel automatically (through OAuth) behind the page's "include my channel" switch; --own none leaves it out.
  2. Publish with the Artifact tool:
    • file_path: artifacts/competitor-board/index.html
    • files: the tool's data.publish_files, as is
    • capabilities: {"db": {}} (the page keeps the competitor list there)
    • icon: radar
  3. Seed the page's list with ArtifactData batch, one set per channel the tool returned without error: collection competitors, doc_id = the handle lowercased without @, data {"handle": "@handle", "enabled": true, "addedAt": "<ISO time>"}. Skip the channel whose own is true: it is not a competitor.
  4. Write memory/competitor_board.md with the URL and the date, and add its line to memory/MEMORY.md.

Sync

  1. ArtifactData list of collection competitors (page with cursor until there is no next_cursor). Save the documents' data objects as a JSON array in the scratchpad. Disabled ones stay in: the user may switch them back on.
  2. python3 tools/competitor_board.py --handles-file <that file>
  3. Publish with Artifact: the board's url, file_path artifacts/competitor-board/index.html, files = data.publish_files as is (it already carries null for channels that were removed). Omit capabilities: the stored declaration carries forward.
  4. Any channel with error means the @ does not exist. Tell the user, and check for look-alikes: a handle can be squatted by an empty channel (e.g. a 2-subscriber account holding the obvious name while the real creator uses a variant). Resolve the real one with the channel ID from config/channels_lists.json if the creator is there.

Monthly plan tab

The page also holds an editable sheet of the videos planned per month, 8 per month by default (2 per pillar, for a 2-a-week cadence). The user edits it on the page; you write ideas into it with ArtifactData:

  • settings/pillars: {"items": [{"id": "p1", "name": "..."}, ...]}. Take the names from memory/; never invent a pillar the user has not named.
  • plan/<YYYY-MM>-<slot>: month, slot, pillar (an id above; pillar names are one or two words), title, pattern, ref_id / ref_title / ref_channel / ref_ratio / ref_views (the main backing outlier, verbatim from the board data), format_refs (format references, shown as links: [{"id", "title", "channel", "ratio", "views"}], verbatim from the board data; the user can add and remove them on the page), tam_views (the topic's tam_views_6m from topic_tam.py), status (index string: 0 idea … 4 published, 5 dropped).
  • views_floor / views_mid / views_ceiling: expected views from python3 tools/topic_tam.py --topic "Name::regex" (a heuristic estimate: write the topic, its market ratio and the user's own record on it into views_basis, shown when hovering the expected views). When the user has 2 or more videos on the topic, their own median beats the market projection. views_real is filled after publishing.
  • Read the rows before writing and pin if_version: the user may have edited them. Never overwrite a title or note the user wrote.

Reading the board (before quoting any figure from it)

  • ratio is against the channel's own median, so a small channel can post 100x with modest views. Quote the views next to the ratio.
  • ad_suspect videos are left out of the median and hidden by default. An outlier with collapsed likes per 1k views is bought reach, not a format that worked: never recommend copying it.
  • Videos under about 7 days old carry a deflated ratio; 1-4 week old videos run high on growing channels. Say so when the top of the list is that young.
  • Fewer than 15 videos in a channel's window makes its median unstable.

SOURCES

competitor_board.py is derived (the ratio and views per hour are computed over API data). ad_suspect is heuristic: declare it as an estimate, never as YouTube data. paid is youtube_api (the creator's own declaration) and marks a sponsorship, not bought views. The competitor list on the page is user data, not a measurement.

OUTPUT

The board's link, plus what changed in one short block: channels added, removed, not found, and how many videos and likely ads the sync produced. If asked what stands out, the top organic outliers from data.top_organic_outliers with title, channel, ratio and views verbatim from the tool.

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.