Pp Youtube

<!-- GENERATED FILE — DO NOT EDIT.
     This file is a verbatim mirror of library/media-and-entertainment/youtube/SKILL.md,
     regenerated post-merge by tools/generate-skills/. Hand-edits here are
     silently overwritten on the next regen. Edit the library/ source instead.
     See the repository agent guide, section "Generated artifacts: registry.json, cli-skills/". -->

YouTube — Printing Press CLI

Prerequisites: Install the CLI

This skill drives the youtube-pp-cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first:

1. Install via the Printing Press installer. It defaults binaries to $HOME/.local/bin on macOS/Linux and %LOCALAPPDATA%\Programs\PrintingPress\bin on Windows:

   npx -y @mvanhorn/printing-press-library install youtube --cli-only

2. Verify: youtube-pp-cli --version
3. Ensure the reported install directory is on $PATH for the agent/runtime that will invoke this skill.

If the npx install fails (no Node, offline, etc.), fall back to a direct Go install (requires Go 1.26.6 or newer). This installs into $GOPATH/bin (default $HOME/go/bin), so add that directory to $PATH instead:

go install github.com/mvanhorn/printing-press-library/library/media-and-entertainment/youtube/cmd/youtube-pp-cli@latest

If --version reports "command not found" after install, the runtime cannot see the binary directory on $PATH. Do not proceed with skill commands until verification succeeds.

The YouTube Data API v3 read surfaces that matter for market and competitor research with complete parameter wiring, feeding a local SQLite databank designed for competitor monitoring: watch your ~15 competitors, monitor refreshes them for ~20-40 quota units per run, and velocity, growth, breakouts, comments-mine, and packaging turn the accumulated snapshots into current market intelligence that lagging analytics platforms deliver one to two weeks late.

When to Use This CLI

Reach for this CLI whenever the task is YouTube market or competitor research: tracking a fixed set of competitor channels over time, measuring what is gaining views right now, discovering fresh breakout videos in a niche, mining comments for audience signal, or collecting titles, thumbnails, and hooks for packaging analysis. It is the right tool when the answer should come from a locally owned, regularly refreshed databank instead of a lagging external analytics platform.

Anti-triggers

Do not use this CLI for:

• Do not use this CLI for your own channel's private analytics (revenue, retention, demographics, traffic sources) - that is the OAuth-only YouTube Analytics API, which this CLI deliberately excludes
• Do not use it to upload, edit, rate, or delete videos or manage a channel - all write operations are out of scope
• Do not use it to download video or audio media - use yt-dlp for media files
• Do not use it for wide-market computed judgments like cross-niche outlier scores or monetization estimates - curated analytics databases own that; this CLI owns fresh data on the channels you track

Unique Capabilities

These capabilities aren't available in any other tool for this API.

Competitor monitoring machine

• watch — Register the competitor channels your monitoring machine tracks, in a typed watchlist table you own.

  Defines the tracked market once; every later monitoring command runs against it without re-specifying channels.

  youtube-pp-cli watch add @mkbhd --json

• monitor — Refresh every watched channel in one run: stats snapshot, new uploads, re-snapshot of recent video statistics.

  One command keeps the databank current, so market answers are hours old instead of weeks old.

  youtube-pp-cli monitor --json

• velocity — See which tracked videos are gaining views fastest right now, computed from real between-snapshot deltas.

  Current market movement - what is taking off today, not what took off two weeks ago.

  youtube-pp-cli velocity --json

• growth — Channel-level subscriber, view, and upload-count deltas between dated local snapshots.

  Tells an agent whether a competitor is accelerating without any external history service.

  youtube-pp-cli growth @mkbhd --json

• backfill — Pull a channel's complete upload history with statistics into the local databank in one command.

  Run once per competitor; every later question about that channel is answered offline for free.

  youtube-pp-cli backfill @mkbhd --json

• workspace — Named databanks: keep the competitor machine in one database and explore a new niche in another, switching instantly.

  Lets an agent spin up a clean research sandbox per niche without risking the production watchlist databank.

  youtube-pp-cli workspace list --json

• auth keys — Store multiple named YouTube API keys, switch between them instantly, and optionally fail over automatically via --rotate when one runs out of quota.

  An agent can finish large collection jobs without human intervention when the first key's daily quota is spent.

  youtube-pp-cli auth keys list --json

Fresh market discovery

• breakouts — Chain search filters into a matrix (terms x upload window x duration x region), join results to channel size, and rank fresh high-momentum videos.

  Finds niche breakouts days after upload, weeks before they reach lagging analytics platforms.

  youtube-pp-cli breakouts "berlin history" --days 14 --json

• comments-mine — Sync comments into a typed full-text-searchable table and report top-liked comments, keyword frequencies, and audience questions.

  Fast audience signal from data you own - what viewers praise, ask, and complain about across a channel.

  youtube-pp-cli comments-mine @mkbhd --json

• packaging — Collect titles, thumbnails (downloaded as local image files), and hook text from transcript openings into a packaging table.

  Hands a multimodal agent everything it needs for thumbnail and hook analysis without any scraping or manual collection.

  youtube-pp-cli packaging @mkbhd --json

Command Reference

youtube — YouTube Data API v3 (read-only, api-key) for market and competitor analysis

• youtube-pp-cli youtube activities-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube captions-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube channel-sections-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube channels-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube comment-threads-list — Retrieves a list of top-level comment threads, filterable by video, channel, or thread id.
• youtube-pp-cli youtube i18n-languages-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube i18n-regions-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube playlist-items-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube playlists-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube search-list — Retrieves a list of search resources
• youtube-pp-cli youtube video-categories-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube videos-list — Retrieves a list of resources, possibly filtered.
• youtube-pp-cli youtube channel-uploads — List a channel's most recent uploads (resolves @handle or channelId, then walks the uploads playlist).
• youtube-pp-cli youtube playlist-enrich — Resolve a playlist to per-video metadata + transcript + description in one concurrent call.
• youtube-pp-cli youtube search-bulk — Search YouTube for multiple terms in one call, return top-N per term.
• youtube-pp-cli youtube videos-comments — Fetch top comments for a video, ranked by like count across pages.
• youtube-pp-cli youtube videos-embed — Print embed HTML, iframe, or markdown snippet for a video.
• youtube-pp-cli youtube videos-enrich — One video's metadata + transcript + description in one call.
• youtube-pp-cli youtube videos-links — Extract resource links from a video description (expands short links, skips social noise).
• youtube-pp-cli youtube videos-related — Find related videos, shared-topic ranking above same-channel.
• youtube-pp-cli youtube videos-transcript — Fetch the transcript without OAuth (timedtext; --format markdown|text|json).

Finding the right command

When you know what you want to do but not which command does it, ask the CLI directly:

youtube-pp-cli which "<capability in your own words>"

which resolves a natural-language capability query to the best matching command from this CLI's curated feature index. Exit code 0 means at least one match; exit code 2 means no confident match — fall back to --help or use a narrower query.

Recipes

Stand up the monitoring machine

youtube-pp-cli watch add @mkbhd --json

Add each competitor once; backfill seeds history and every monitor run keeps them current

What moved today

youtube-pp-cli velocity --agent --select items.title,items.views_per_day

Between-snapshot view velocity for tracked videos, narrowed to the fields an agent needs

Fresh breakouts in a niche

youtube-pp-cli breakouts "berlin history" --days 14 --json

Chained filter matrix joined to channel size - high views-per-subscriber uploads from the last two weeks

Packaging dossier for the agent

youtube-pp-cli packaging @mkbhd --json

Titles, local thumbnail files, and hook text side by side, ready for multimodal packaging analysis

What the audience keeps asking

youtube-pp-cli comments-mine @mkbhd --json

Top-liked comments, keyword frequencies, and extracted questions from the synced comment table

Auth Setup

Set YOUTUBE_API_KEY to a YouTube Data API v3 key (create one at console.cloud.google.com under APIs & Services > Credentials), or store it once with auth set-token. A key in the environment overrides the stored one - if doctor shows auth_source env and calls fail with HTTP 400 'API key not valid', the environment copy is stale: unset it or update it. Read-only public-data operations only; no OAuth anywhere.

Run youtube-pp-cli doctor to verify setup.

Agent Mode

Add --agent to any command. Expands to: --json --compact --no-input --no-color.

• Pipeable — JSON on stdout, errors on stderr

• Filterable — --select keeps a subset of fields. Dotted paths descend into nested structures; arrays traverse element-wise. Critical for keeping context small on verbose APIs:

  youtube-pp-cli youtube activities-list --part snippet --agent --select contentDetails,etag,id

• Previewable — --dry-run shows the request without sending

• Offline-friendly — sync/search commands can use the local SQLite store when available

• Non-interactive — never prompts, every input is a flag

• Read-only — do not use this CLI for create, update, delete, publish, comment, upvote, invite, order, send, or other mutating requests

Response envelope

Commands that read from the local store or the API wrap output in a provenance envelope:

{
  "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."},
  "results": <data>
}

Parse .results for data and .meta.source to know whether it's live or local. A human-readable N results (live) summary is printed to stderr only when stdout is a terminal AND no machine-format flag (--json, --csv, --compact, --quiet, --plain, --select) is set — piped/agent consumers and explicit-format runs get pure JSON on stdout.

Paths and state

Agents should treat the CLI's path resolver as part of the runtime contract:

• Use --home <dir> for one invocation, or set YOUTUBE_HOME=<dir> to relocate all four path kinds under one root.

• Use per-kind env vars only when a specific kind must diverge: YOUTUBE_CONFIG_DIR, YOUTUBE_DATA_DIR, YOUTUBE_STATE_DIR, YOUTUBE_CACHE_DIR.

• Resolution order is per-kind env var, --home, YOUTUBE_HOME, XDG (XDG_CONFIG_HOME, XDG_DATA_HOME, XDG_STATE_HOME, XDG_CACHE_HOME), then platform defaults.

• config contains settings like config.toml and profiles. data contains credentials.toml, data.db, cookies, and auth sidecars. state contains persisted queries, jobs, and teach.log. cache contains regenerable HTTP/cache files.

• Stored secrets live in credentials.toml under the data dir. Existing legacy config.toml secrets are read for compatibility and leave config.toml on the first auth write.

• Run youtube-pp-cli doctor --fail-on warn to surface path and credential-location warnings. agent-context exposes a schema v4 paths block for agents that need the resolved dirs.

• For MCP, pass relocation through the MCP host config. The MCP binary does not inherit CLI flags:

  {
    "mcpServers": {
      "youtube": {
        "command": "youtube-pp-mcp",
        "env": {
          "YOUTUBE_HOME": "/srv/youtube"
        }
      }
    }
  }

⚠️ Two files deliberately live OUTSIDE the relocatable tree, in the platform config dir (~/Library/Application Support/youtube-pp-cli/ on macOS): workspaces.json (the workspace registry must sit outside workspace homes or switching becomes self-referential) and keyring.json (quota is per key, not per workspace). Consequence: --home/YOUTUBE_HOME does NOT isolate the key ring or workspace registry — keys add/use and workspace create/use mutate shared state even under an isolated home.

Fleet precedence: an inherited per-kind env var overrides an explicit --home for that kind. Use YOUTUBE_HOME or per-kind vars as durable fleet levers, and use --home only for a single invocation. Relocation is not reversible by unsetting env vars; move files manually before clearing YOUTUBE_HOME, or doctor will not find credentials left under the former root.

The analyst databank (SQL schema)

Every analyst command writes into one local SQLite databank — the file doctor --json reports as the store path. The filename is scoped by the active API key (data-<hash>.db); workspace switches between entirely separate databank files. Agents query it two ways: the MCP sql tool (read-only, validated) and the MCP search / CLI search full-text surface.

monitor, backfill, breakouts, comments-mine, and packaging feed these tables automatically (write-through); velocity and growth are computed from consecutive yt_video_snapshots / yt_channel_snapshots rows — two runs on different days are the minimum for a non-empty answer.

Example queries (all verified against a live-populated store):

-- What moved: views per video from the latest snapshots
SELECT s.video_id, v.title, s.view_count, s.captured_at
FROM yt_video_snapshots s JOIN yt_videos v USING(video_id)
ORDER BY s.captured_at DESC, s.view_count DESC LIMIT 20;

-- Audience signal: most-liked comments mentioning a topic (FTS5)
SELECT c.like_count, c.author, c.text
FROM yt_comments_fts f JOIN yt_comments c ON c.rowid = f.rowid
WHERE yt_comments_fts MATCH 'gemini' ORDER BY c.like_count DESC LIMIT 10;

-- Growth rate per tracked channel between first and last snapshot
SELECT channel_id,
       MAX(subscriber_count) - MIN(subscriber_count) AS subs_delta,
       MIN(captured_at) AS first_seen, MAX(captured_at) AS last_seen
FROM yt_channel_snapshots GROUP BY channel_id;

Automatic learning

This CLI ships a self-capturing learning loop. The CLI does its own bookkeeping: every invocation is journaled locally, a failed flag followed by a corrected retry auto-derives a flag_alias candidate, and a teach on a query family without a playbook auto-synthesizes a playbook_candidate from the session's journal. Your job is judgment only: recall first, act on surfaced candidates, teach the final answer, playbook amend when you observe a correction. You never record failures by hand.

Step 1: recall before any discovery

Before list/search/drill commands on a new user question, run:

youtube-pp-cli recall "<user's question>" --agent

The response envelope:

{
  "query": "...",
  "normalized": "<normalized form>",
  "query_entities": ["..."],
  "found": true | false,
  "match_score": 0.0,
  "results": [
    { "resource_id": "...", "resource_type": "...", "venue": "...",
      "confidence": 2, "entity_match": "exact|partial|unknown",
      "source": "taught|preseed|pattern", "warnings": ["..."] }
  ],
  "mismatches": [ /* only when --debug-mismatches */ ],
  "warnings": [ /* top-level */ ],
  "candidates": [
    { "id": 12, "class": "flag_alias | playbook_candidate",
      "summary": "...", "sightings": 3, "last_seen": "...",
      "rationale": "...",
      "next_action": ["<trial command>", "youtube-pp-cli learnings confirm 12"] }
  ],
  "playbook": {
    "query_family": "...",
    "playbook": {
      "steps": [ { "cmd": "<command with {slot} substitution>", "purpose": "..." } ],
      "entity_slots": ["$ENTITY"],
      "expected_tool_calls": 3
    },
    "slots_resolved": { "$ENTITY": { "token": "<live token>", "canonical": "<canonical>" } },
    "notes": "<workarounds + gotchas for this query family>"
  },
  "notes": "<duplicate surface for non-playbook callers>"
}

Empty-store short-circuit: if the store has no learnings, playbooks, or candidates yet (recall finds nothing and learnings list and learnings candidates are both empty), skip recall for the rest of this session instead of taxing every query; resume recall-first once something has been taught.

Step 2: decision tree

Read candidates, playbook, notes, results[0], and warnings in that order:

if Candidates present (warnings include "candidates_present"):
    -> candidates are try-then-confirm, never facts. Follow each candidate's
       two-step next_action verbatim: run the trial command first, then run
       `learnings confirm <id>` only after the trial verified the behavior.
       Reject a wrong candidate with `learnings reject <id>`.
    -> NEVER re-teach something recall surfaced as a candidate; confirm or
       reject that candidate instead of teaching a duplicate.
    -> candidates ride alongside playbooks and resource hits, not instead of
       them; continue with the branches below after acting on them.

if Playbook present:
    -> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose)
    -> replay Playbook.steps in order, substituting Playbook.slots_resolved entries
       for the entity slot tokens. If a step's slot is unresolved, fall back to
       discovery for that step only.
    -> the Playbook's expected_tool_calls is a budget; if you find yourself running
       materially more, record the divergence via `youtube-pp-cli playbook amend`
       at end-of-session.

elif Notes present (no Playbook):
    -> read Notes verbatim before any discovery step; they carry known gotchas
       for this query family even when no structured choreography exists yet.

elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2:
    -> skip discovery; fetch live data for Results[*].ResourceID in parallel

elif Found AND Results[0].EntityMatch == "partial":
    -> candidate hint, NOT a hit; read the resource title to validate before trusting

elif (any row in Mismatches[] when --debug-mismatches was passed):
    -> treat as cold start; the stored learning is for a different entity
       (different canonical resolved from query_entities)

else:  // Found == false, no playbook, no notes
    -> cold start; run discovery normally; teach the answer afterward (Step 4).
       If the family has no playbook yet, that teach auto-synthesizes a
       playbook candidate from this session's journal - you do not need to
       record one by hand.

Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a Results[] hit - use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping mismatches; pass --debug-mismatches only when investigating cold-start surprises.

Candidate judgment details: learnings confirm <id> prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. learnings reject <id> tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; youtube-pp-cli learnings candidates lists the full open set.

Graceful degradation: if learnings confirm is an unknown command, you are driving an older binary - ignore the candidates guidance and follow the rest of the protocol.