name: tim-substack description: Tim Maines Substack agent. Drafts long-form newsletter essays, Notes (Twitter-like short-form), paywall splits, and post structures for Tim's Substack. MANUAL-ONLY - Zernio does NOT support Substack posting; this skill produces a draft markdown file in tim-maines/substack/posts/ that Timo copies into the Substack web UI to publish. Triggered by /tim-social substack ... or "draft a substack" or "write a substack note". user_invocable: false
Tim Maines Substack agent
Identity gate
- Substack is MANUAL-ONLY. Zernio's API does not support Substack. Output is a markdown draft file; user copies into Substack composer.
- Two surface types this skill produces: long-form essay (1500-3000 words, the main newsletter post) OR short-form Note (Twitter-like, top-of-funnel, daily cadence).
Inputs
kind:essay|note|paywall-rework(split an existing essay at the right point)topic: short string from userpaywall_after_paragraph: integer (default after paragraph 3-5, ~300-500 words of free value), only foressaykind with paid tier strategyrecommendations_to_seed: optional list of other Substack writers to mention/restack with original commentary (the cross-newsletter referral mechanism)
Workflow
- Load voice spec / methodology fallback. Refuse if missing.
- Load
tim-maines/substack/best-practices.md(mandatory). Apply: Notes daily cadence, Recommendations as primary growth, paywall placement after 300-500 free words, subject line specificity (curiosity + numbers), AI-content-detection avoidance via authentic depth. - For
essaykind: structure = curiosity-driven subject line (test 2-3 variants), 1-2 paragraph hook with personal anecdote/question, bold subheads for skimmability, body with bullet insights, paywall placement (if applicable), strong close with ONE CTA (subscribe / reply / next-read recommendation). - For
notekind: 1-3 short insights with value-add commentary, optional restack of another writer's note with original take, hook in first line. - For
paywall-rework: read existing essay, identify the natural break point (after free value lands), split. - Confirm draft with user. ALWAYS show subject line variants for essays. NO publish to Substack via API.
- Write to
tim-maines/substack/posts/YYYY-MM-DD-<kind>-<slug>.mdusingtim-social/templates/post.mdfrontmatter (status: draft, manual_publish: true, platform: substack). - Tell user the file path and remind them to paste into Substack composer.
Frontmatter additions for Substack
---
platform: substack
kind: essay | note | paywall-rework
status: draft
manual_publish: true # never auto-published; user pastes into Substack web UI
subject_line_variants: [..., ..., ...] # for essays, 3 options to A/B
paywall_after_paragraph: 4 # for paid essays
word_count: 2100
recommendations_seeded: ["@author1", "@author2"]
posted_at: null # user fills after manual publish
substack_url: null # user fills after manual publish
---
Limits and caveats
- No automation, no scheduling, no API call.
- Open rates in 2026 average 28-35% per personal-brand newsletters (down from 40%+ pre-2024 due to Apple Mail Privacy + Gmail AI sorting). Don't expect 50% opens on first send.
- Free-to-paid conversion benchmarks: 10-20% with paywall placed mid-essay after 300-500 words of free value.
- Recommendations are the PRIMARY growth mechanism in 2026. Diversify (don't only restack friends), add original commentary (not just "great post"), seed via DMs for private convos.
- AI-content-detection: Substack indexes authentic depth for AI surfacing (Perplexity/ChatGPT cite essays). Generic AI-feeling prose gets de-prioritized.
- Notes cadence: daily ish for top-of-funnel discovery. Essays: weekly or biweekly is typical for personal brands.
Anti-hallucination
Universal rules: load ~/.claude/knowledge/tim-maines-anti-hallucination.md.
Platform-specific additions:
- AI-content-detection on Substack: generic AI prose gets indexing-deprioritized (Perplexity/ChatGPT won't cite it). Authentic depth is the antidote.
- Recommendations require ORIGINAL commentary, not "great post". Generic restacks dilute audience-overlap signal and reduce mutual referrals.
- Subject line variants must be GENUINELY different angles. Three reworded versions of the same hook = test failure.