skillpacks 0.1.1 → 0.1.2

package/packs/youtube-ops/claude/youtube-peer-benchmark/archive/v0.4/SKILL.md

@@ -0,0 +1,384 @@
+---
+name: youtube-peer-benchmark
+description: Benchmark a YouTube channel against peers — discover comparable creators, pull real metrics via yt-dlp, compare performance at equivalent stages, and diagnose whether the gap is quality, volume, format, or conversion
+type: research
+version: v0.4
+argument-hint: "<channel URL or handle> [--niche 'keyword phrase'] [--peers @handle1,@handle2] [--count N]"
+interview_depth: none
+visual_tier: visual
+---
+
+## Pack Availability Guard
+
+Before telling the user to run a skill from another project-local pack, check `.agents/project.json.enabled_packs`. If the target pack is not enabled, recommend `/pack install <pack>` instead of the target skill. Global skills are always valid. Skills from this same pack are valid because the current skill is already running from that pack.
+
+# YouTube Peer Benchmark
+
+Invoke as `/youtube-peer-benchmark`.
+
+## Report-First Approval Gate
+
+Default to scope-first approval: before synthesized research, inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions in a `review` alignment page plus a concise conversation summary.
+
+Do not perform synthesized research, rank candidates, make recommendations, or write working packets or canonical deliverables until final compiled YAML approves the research scope. Minimal pre-approval discovery may identify available files, source categories, and open questions; label it as scope evidence, not findings.
+
+After approved research-scope YAML, perform the research and write only the non-canonical working packet defined in the staged workflow. Then update the `review` alignment page with findings and stop again for feedback-only YAML or final compiled YAML artifact approval before creating or updating canonical research, spec, or task files.
+
+Do not include `Recommended next skill`, `Recommended next command`, or downstream routing language. The approval request itself is the next action. Only emit next-skill routing after the approved artifact has been written or updated.
+
+## Staged Research Workflow
+
+Use this staged workflow for synthesized research or report outputs that would create or update canonical research, spec, or task files.
+
+1. **Stage 1 - Scope discovery and approval.** Inspect only enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` HTML alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions/confidence, proposed working-packet and canonical output paths, and research-scope approval gates. Stop for final compiled YAML approval of the research scope. Do not perform synthesized research, rank candidates, make recommendations, or write working packets, canonical research, spec, or task files in Stage 1.
+2. **Stage 2 - Research and artifact review.** Only after approved research-scope YAML with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback, perform the synthesized research, run required source/code checks, and write only a non-canonical working packet: flat mode uses `research/_working/preliminary-<skill>-research.md`; product-path mode uses `research/{slug}/_working/preliminary-<skill>-research.md`. Replace `<skill>` with this skill's `name` value. Raw evidence or search logs may remain as supporting evidence where this skill already requires them, but synthesized deliverables stay in the working packet. Update the `review` HTML alignment page with the full preliminary packet, evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Stop for either feedback-only YAML or final compiled YAML. Feedback-only YAML revises the working packet and page, then remains in Stage 2.
+3. **Stage 3 - Finalize approved artifacts.** Consume final compiled YAML for artifact approval only when it has no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths below, and convert the alignment page to `confirmed` with the approval record preserved.
+
+Canonical output paths remain unchanged. Search logs and other supporting evidence remain allowed only where this skill's output contract already requires them.
+
+## Evidence And Feedback Handling
+
+Treat user feedback as input to evaluate, not as automatic ground truth.
+
+- For factual, evidentiary, technical, or source-backed claims: verify against available evidence. If the user appears to misunderstand the evidence or states something factually incorrect, push back clearly and cite the evidence. Do not rewrite findings merely to agree.
+- For taste, brand, positioning preference, risk appetite, prioritization, or other subjective judgment calls: weigh user feedback heavily and adapt the recommendation unless it conflicts with verified evidence.
+- When feedback mixes facts and preference, separate them explicitly: correct the factual part, then incorporate the preference where it is a legitimate judgment call.
+- When uncertain, say what is known, what is inferred, and what would change the conclusion.
+
+## Prerequisites
+
+- **yt-dlp**: Required. Install: `brew install yt-dlp` or `pip install yt-dlp`.
+- **Python 3**: Required for data analysis. No special packages needed beyond stdlib.
+- **Browser cookies** (optional): If yt-dlp returns incomplete data (NA values for views/likes), retry with `--cookies-from-browser brave` (or chrome/firefox/safari). Mention this to the user if needed.
+
+Check `command -v yt-dlp` before proceeding. If missing, tell the user what to install and stop.
+
+## Process
+
+### 1. Parse Arguments
+
+- `$ARGUMENTS` must contain a YouTube channel URL or handle.
+- Optional `--niche 'keyword phrase'` overrides automatic niche detection (e.g. `--niche 'nextjs tutorial'`).
+- Optional `--peers @handle1,@handle2,...` specifies known competitors to include (in addition to discovered ones).
+- Optional `--count N` limits the target channel to the N most recent videos (default: all).
+- Normalize handles: `@handle` → `https://www.youtube.com/@handle/videos`.
+
+### 2. Gather Target Channel Baseline
+
+If a current `/youtube-channel-audit` exists for the target channel, read raw metadata from `research/youtube/data/<slug>/videos-*.jsonl`. Otherwise, fetch directly.
+
+Pull full metadata for the target channel:
+
+```bash
+yt-dlp --cookies-from-browser brave --skip-download \
+  --print "%(id)s|%(title)s|%(upload_date)s|%(view_count)s|%(like_count)s|%(comment_count)s|%(duration)s|%(channel_follower_count)s" \
+  "CHANNEL_URL/videos"
+```
+
+If `--count N` is specified, add `--playlist-end N`.
+
+From this, compute the target channel profile:
+
+| Metric | How |
+|--------|-----|
+| Subscriber count | From `channel_follower_count` |
+| Total videos | Count of results |
+| Total views | Sum of `view_count` |
+| Channel age | First upload date to today |
+| Publishing cadence | Videos per month (total videos / months active) |
+| Avg views/video | Total views / total videos |
+| Median views | Sorted middle value |
+| Top video | Highest view count |
+| Floor video | Lowest view count |
+| Videos under 100 views | Count |
+| Videos over 1K views | Count |
+| Like rate | Total likes / total views |
+| View-to-sub ratio | Total views / subscribers |
+| Sub-per-view | Subscribers / total views (conversion efficiency) |
+| Publishing gaps | All gaps > 14 days between consecutive uploads |
+| Content categories | Classify each video by format: tutorial, commentary, vod, interview, meta, shorts, etc. |
+
+Also compute:
+- Performance by category: views, likes, comments, count, avg views per category
+- Performance by time period (quarterly): videos published, total views, avg views per quarter
+
+### 3. Discover Peer Channels
+
+#### Automatic Discovery
+
+Determine the target channel's niche from its top-performing video titles and tags. Use the 3-5 most distinctive keywords (not generic terms like "tutorial" or "how to").
+
+Search YouTube for creators making similar content:
+
+```bash
+yt-dlp --cookies-from-browser brave --skip-download \
+  --print "%(channel)s|%(channel_id)s|%(channel_follower_count)s|%(view_count)s|%(upload_date)s|%(title).80s" \
+  "ytsearch20:KEYWORD1 KEYWORD2 KEYWORD3 2025"
+```
+
+Run 3-5 searches with different keyword combinations from the target's niche to build a broad candidate list.
+
+#### Filter and Tier Candidates
+
+From search results, extract unique channels. Discard:
+- The target channel itself
+- Channels with no videos in the last 6 months (inactive)
+- Channels that only appeared once across all searches (likely tangential)
+
+Sort remaining channels into tiers:
+
+| Tier | Criteria | Purpose |
+|------|----------|---------|
+| **Actual peers** | Similar subscriber count (0.25x–4x of target) | Direct comparison |
+| **Aspirational peers** | 5x–25x subscribers, same niche | Growth trajectory reference |
+| **Category leaders** | 50x+ subscribers, same niche | Ceiling benchmark (not for direct comparison) |
+
+Include any channels specified via `--peers` in addition to discovered ones, placed in the appropriate tier.
+
+Select up to 3 actual peers, 3 aspirational peers, and 2 category leaders for analysis. Prefer channels with the most keyword overlap with the target.
+
+### 4. Fetch Peer Channel Data
+
+For each selected peer channel, pull metadata:
+
+```bash
+yt-dlp --cookies-from-browser brave --skip-download \
+  --print "%(id)s|%(title)s|%(upload_date)s|%(view_count)s|%(like_count)s|%(comment_count)s|%(duration)s|%(channel_follower_count)s" \
+  "https://www.youtube.com/@HANDLE/videos"
+```
+
+For channels with 50+ videos, fetch the full catalog to enable stage-matched comparison. Use `--playlist-end` only if the channel has 200+ videos, in which case fetch the most recent 100 plus the earliest 50 (to capture both current performance and early-stage trajectory).
+
+Compute the same baseline metrics as step 2 for each peer.
+
+### 5. Stage-Matched Comparison
+
+This is the most important analysis. Do not just compare current stats — compare channels at equivalent stages.
+
+For each aspirational/leader peer, isolate their **first N videos** where N = the target channel's total video count. This answers: "when this peer had the same number of videos, how were they doing?"
+
+Compute for the stage-matched slice:
+
+| Metric | Target | Peer (at same video count) |
+|--------|--------|---------------------------|
+| Total views | | |
+| Avg views/video | | |
+| Median views | | |
+| Best single video | | |
+| Videos under 100 views | | |
+| Videos over 1K views | | |
+| Time span to reach N videos | | |
+| Avg days between videos | | |
+| Format consistency (category count) | | |
+
+### 6. Diagnose the Gap
+
+For each peer comparison, categorize the gap across five dimensions:
+
+#### Volume Gap
+- How many more videos has the peer published in the same calendar time?
+- What is the peer's publishing cadence vs. the target's?
+- How many publishing gaps >14 days does each have?
+
+#### Consistency Gap
+- Does the peer have fewer format categories (more focused)?
+- Does the peer have a higher floor (fewer low-performing videos)?
+- Does the peer maintain cadence without multi-week breaks?
+
+#### Quality Gap (or lack thereof)
+- Compare ceiling: target's best video vs. peer's best at same stage
+- Compare like rates on comparable videos
+- Compare comment engagement
+
+#### Conversion Gap
+- Views-to-subscriber ratio: how efficiently does each channel convert viewers to subscribers?
+- If the target has worse conversion, hypothesize why (format scatter, unclear channel promise, etc.)
+
+#### Format Gap
+- What content types does the peer make vs. the target?
+- Which formats drive the most views for each?
+- Does the target make content types that consistently underperform?
+
+### 7. Write Report
+
+Save to `research/youtube/peer-benchmark-<primary-slug>-YYYY-MM-DD.md`:
+
+Create the `research/youtube/` directory if it does not exist.
+
+```markdown
+# YouTube Peer Benchmark — [Channel Name]
+
+> Channel: [URL]
+> Subscribers: N
+> Total videos: N
+> Date: YYYY-MM-DD
+> Niche: [detected or specified niche keywords]
+
+## Target Channel Baseline
+
+### Overall Stats
+- Subscribers: N
+- Total videos: N
+- Total views: N
+- Channel age: N months
+- Publishing cadence: N videos/month
+- Avg views/video: N
+- Median views: N
+- Top video: N views — [title]
+- Floor: N videos under 100 views
+
+### Performance by Category
+
+| Category | Videos | Total Views | Avg Views | Like Rate | Notes |
+|----------|-------:|------------|----------:|----------:|-------|
+| ... | | | | | |
+
+### Performance by Quarter
+
+| Quarter | Videos | Total Views | Avg Views |
+|---------|-------:|------------|----------:|
+| ... | | | |
+
+### Publishing Gaps (>14 days)
+
+| Gap | From | To |
+|----:|------|------|
+| ... | | |
+
+## Peer Channels
+
+### Tier: Actual Peers (similar size)
+
+| Channel | Subs | Videos | Since | Cadence | View Range |
+|---------|-----:|-------:|-------|---------|------------|
+| ... | | | | | |
+
+### Tier: Aspirational Peers (5-25x)
+
+| Channel | Subs | Videos | Since | Cadence | View Range |
+|---------|-----:|-------:|-------|---------|------------|
+| ... | | | | | |
+
+### Tier: Category Leaders (50x+)
+
+| Channel | Subs | Videos | Since | Cadence | View Range |
+|---------|-----:|-------:|-------|---------|------------|
+| ... | | | | | |
+
+## Stage-Matched Comparisons
+
+### [Peer Name] at [N] videos vs. [Target Name] at [N] videos
+
+| Metric | [Target] | [Peer] (first N) |
+|--------|----------|-------------------|
+| Total views | | |
+| Avg views/video | | |
+| Median views | | |
+| Best single video | | |
+| Videos under 100 views | | |
+| Videos over 1K views | | |
+| Time to publish N videos | | |
+| Avg days between videos | | |
+
+### ...
+
+## Gap Diagnosis
+
+### Volume
+[Is the target underproducing relative to peers? By how much? What does peer cadence look like?]
+
+### Consistency
+[Does the target have more format scatter, more gaps, a lower floor?]
+
+### Quality
+[Is the target's ceiling competitive? Are like rates comparable?]
+
+### Conversion
+[Views-to-sub efficiency compared to peers. Why the difference?]
+
+### Format
+[What content types work for peers that the target isn't making? What is the target making that peers avoid?]
+
+## Verdict
+
+### Where [Channel] is ahead
+1. [Specific advantage with evidence]
+2. ...
+
+### Where [Channel] is behind
+1. [Specific gap with evidence]
+2. ...
+
+### What explains the gap
+[2-3 sentences: is this a quality problem, a volume problem, a format problem, or a conversion problem? Be specific.]
+
+### The single most impactful change
+[One concrete action, grounded in the peer data, that would close the biggest gap]
+```
+
+### 8. Summarize In Thread
+
+After saving the report, output to the user:
+
+- Subscriber count vs. peer range
+- The stage-matched headline (e.g. "Your best video outperforms OrcDev's at the same stage, but you publish 4x less often")
+- Top 2 advantages
+- Top 2 gaps
+- The single most impactful change
+- Path to the full report
+
+## Constraints
+
+- **Real data only**: Every number must come from yt-dlp output. Never estimate, interpolate, or fabricate metrics.
+- **Stage-matched comparisons are mandatory**: Never compare a 34-video channel's total stats against a 245-video channel's total stats as if that's meaningful. Always isolate the peer's equivalent stage.
+- **Tier honestly**: Don't call a 200K-subscriber channel a "peer" to a 1K-subscriber channel. Tier them correctly and label the comparison as aspirational or ceiling benchmark.
+- **Diagnose, don't prescribe content**: The output is a competitive diagnosis, not a content calendar. Say "you publish 4x less often" not "you should make 8 videos per month."
+- **Do not compare channels without comparable evidence windows**: Mark subscriber counts or external metrics as unavailable unless present in fetched metadata.
+- **Do not recommend copying peers**: Translate benchmark evidence into differentiated positioning.
+- **Browser cookies**: Try without cookies first. If yt-dlp returns NA for view counts or subscriber counts, retry with `--cookies-from-browser brave` (or whichever browser the user has). Tell the user which browser you're using.
+- **Rate limiting**: If YouTube returns 429 errors, wait 30 seconds and retry once. If it persists, work with whatever data was successfully fetched and note the gaps.
+- **No API keys required**: yt-dlp works against public data. Do not ask for YouTube Data API keys.
+- **Flat playlist for discovery, full metadata for analysis**: Use `--flat-playlist` when you only need video IDs/titles for discovery. Use full `--skip-download` metadata fetch when you need view counts, likes, dates.
+
+## Approved Artifact Handoff
+
+After an approved synthesized write, explicit write/update mode, or any direct artifact mutation:
+
+- List every created or updated synthesized artifact path in the final response.
+- State the verification performed, such as readback, schema/check command, or why no executable verification applies for a Markdown-only strategy artifact.
+- Check and report the relevant git status for intended artifacts when the project is a git repository. If intended artifacts are modified or untracked, make the next action shipping, committing, or an explicit dirty-artifact handoff before recommending downstream strategy work.
+- Do not imply the research workflow is complete while approved artifacts remain untracked or uncommitted unless the user explicitly asked not to ship.
+- If stopping for approval before writing, the approval request remains the next action; do not include downstream routing.
+
+## Intent-Aware Routing
+
+Before applying the default `## Next-Skill Routing` sequence, classify the user's immediate intent and route to the missing action that best serves that intent:
+
+- Strategy refresh: recommend the missing or stale positioning, programming, portfolio, metrics, or product-media artifact.
+- Recording prep: recommend the missing series spec, script, build proof, walkthrough guide, or validation artifact needed before recording.
+- Upload prep: recommend packaging, title/thumbnail, description, chapters, or final metadata work before broader strategy work.
+- Performance review: recommend metrics, cadence, portfolio, peer benchmark, or owner-analytics export work before new content planning.
+- Owner analytics or private/manual platform evidence: route to an explicit manual/guide handoff instead of inventing unavailable metrics.
+- Dirty intended artifacts: route to shipping/commit/handoff first, not another creator strategy skill.
+
+Use the default next-skill sequence only when no stronger user intent, missing artifact, manual blocker, or dirty-artifact handoff applies.
+
+## Next-Step Routing
+
+After writing the artifact, recommend the next contextual creator-media skill in the final response as `Recommended next skill: <command>`.
+
+Default recommendation: `/youtube-search-positioning`.
+
+If the default successor already exists and is current, recommend the first missing or stale downstream creator-media artifact in this order:
+
+`/youtube-channel-audit` -> `/youtube-video-audit` -> `/youtube-vid-research` -> `/youtube-concept-research` -> `/youtube-competitive-research` -> `/youtube-title-thumbnail-audit` -> `/youtube-description-optimizer` -> `/youtube-portfolio` -> `/youtube-peer-benchmark` -> `/youtube-search-positioning` -> `/youtube-cadence-diagnosis` -> `/creator-positioning` -> `/content-programming` -> `/series-spec` -> `/product-led-media-map` -> `/creator-metrics-review`
+
+If the sequence is ambiguous, multiple upstream artifacts are stale, or the recommendation depends on channel-level strategy vs programming-level changes, recommend `/creator-metrics-review` when metrics evidence exists, otherwise recommend the default successor and explain the missing artifact.
+
+## Alignment Page
+
+When this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page following `ALIGNMENT-PAGE.md` in this skill's directory. Output: `alignment/youtube-peer-benchmark-{topic}.html`.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.