Open-source social media skills and automation research
Reviewed: 2026-10-08. Goal: collect the strongest ideas from the existing skill ecosystem into an original, secure, executable Claude Code workflow, without blindly copying code or assuming every README claim is true. This is a researched shortlist plus one 106-skill corpus, not an exhaustive census of every project on GitHub. Sources change; verify their latest release/license before reuse.
Comparative landscape
Section titled “Comparative landscape”| Source | What it contributes conceptually | Opportunity or remaining dependency addressed by ReachWeaver | Code/license decision |
|---|---|---|---|
| social-media-skills/skills | 106 skill modules, brand voice, audience research, hook-writing, visual/video, platform playbooks, campaign calendar, community, publishing and reporting | Combine the workflows with a persistent reviewed outbox, receipts and provenance-aware measurement rather than relying on writing guidance alone | Upstream README says MIT. We independently authored instructions and CLI rather than bulk-copying its prose |
| scrollmark/social-skills | Reads tone/subtext, short-form narrative and video production grammars | Add substantive technical proof, approval, tracking and cross-platform feedback | Inspiration, no code copied |
| namddp/social-media-agent-skills | Cross-platform voice, source mining, drafts and local validation; explicitly no unattended publishing | Provide optional permitted API execution in addition to offline lint and review | Inspiration, no code copied |
| guyaga/claude-code-social-media-skill | Claude skill driving many platforms via Upload Post API | Keep execution trace and local safety independent of a paid third-party posting gateway | Do not copy, optional future adapter |
| publora/skills | 10-network MCP publishing, managed account authorization and scheduling | Separate campaign planning/analytics and independent durable records; document hosted service dependency | Do not copy, optional MCP integration |
| Schmedu/social-post-skill | Brand-aware draft / schedule / edit / revert commands mapped to Social Media Manager MCP | Build explicit local approval gates and distinguish real publish receipts from draft objects | Upstream reports MIT; no code copied |
| Batsirai/social-media-agent | Buffer/Postiz scheduling adapter and content calendar | Treat providers as swappable; avoid assuming a free plan or access to every channel | No code copied; pricing/access recheck needed |
| gitroomhq/postiz-app | Rich self-hosted scheduler, media library, analytics, API/MCP/CLI and team collaboration | Prefer integration over reimplementing full social suite; keep strategy/provenance outside backend | AGPL-3.0. Do not paste internal implementation into this Apache-2.0 repository |
| borghei/Claude-Skills social-media-manager | B2B SaaS calendars, pillars, community routines, crisis planning | Add deterministic actions and complete auditability; avoid inflexible frequency benchmarks | File declares MIT + Commons Clause; do not reproduce its text |
| borghei/Claude-Skills social-content | Platform-native writing, hook patterns and repurposing | Source-backed proofs, measurement link and brand separation | File declares MIT; no code copied |
| xcrrr/claude-skills social-media | Writing for threads, LinkedIn and Instagram | Add operational security, scheduling and feedback loop | File declares MIT; no code copied |
| alirezarezvani/claude-skills social-media-manager | Basic calendar, content pillars, growth analytics | Treat suggested growth thresholds as hypotheses, not universal cross-platform targets | Inspiration, no code copied |
| claude-office-skills/skills twitter automation | X scheduling, threads, analytics tasks | Treat unsupported tool names, unverified posting-time advice and rigid cadence as not-yet-verified; enforce execution checks | Inspiration, no code copied |
Important distinction: The last column describes our implementation/reuse decision; it does not declare that every upstream project has security bugs. Comparisons are against a multi-account Claude-driven system with the user’s specific requirements.
Comprehensive 106-skill source catalog
Section titled “Comprehensive 106-skill source catalog”The following names were enumerated from the source’s GitHub skills/ directory on 2026-10-08 using the GitHub contents API. Names are catalog entries, not 106 bundled/vendored implementations and not endorsements of their current technical correctness. They are indexed here so future maintainers can selectively research specialists rather than blindly install everything in a privileged runtime.
Foundation, discovery, research, planning:
brand-profile, voice-builder, writing-style-and-tone, audience-research, social-strategy, content-pillars, goals-and-kpis, profile-optimization, idea-generation-and-ideation, content-research-and-sourcing, competitor-analysis, viral-reverse-engineering, content-calendar, batch-content-plan, campaign-and-launch-planning, seasonal-and-moment-marketing, trend-jacking, data-and-original-research.
Writing, creative types, storytelling:
hook-writer, caption-writer, linkedin-post-writer, thread-writer, threads-post, text-post-and-microblog, carousel-writer, story-writer, short-form-video-script, reels-script, tiktok-script, scripting-and-storyboarding, reply-and-comment-writer, educational-content-and-how-to, storytelling-and-narrative, contrarian-and-opinion, behind-the-scenes-and-founder, before-after-and-transformation, social-proof-and-testimonials, listicle-and-roundup, meme-and-culture, interactive-content, livestream-and-realtime, podcast-and-audiograms, email-and-newsletter.
Platform specializations:
instagram-growth, instagram-seo, instagram-reels-publishing, tiktok-growth, tiktok-photo-mode, tiktok-video-publishing, linkedin-growth, linkedin-company-pages, x-growth, threads-growth, facebook-strategy, facebook-groups, pinterest-growth, pinterest-seo, pinterest-pin-design, youtube-long-form, youtube-shorts, youtube-publishing-and-metadata, thumbnail-design, reddit-marketing.
Visual, audio and production:
design-and-templates, image-prompt, quote-cards-and-text-graphics, infographic-and-data-viz, ai-image-editing, ai-video, ai-voiceover, ai-music-and-sound, canva, capcut, captions-and-clipping, descript, flux, heygen, ideogram, kling, luma, nano-banana, opus-clip, runway, suno, synthesia, talking-head-and-piece-to-camera, veo-3.
Distribution, operations, growth, measurement:
hashtag-strategy, social-seo, ai-search-optimization, cross-platform-repurposing, content-recycling, link-in-bio-and-traffic, lead-magnets-and-funnels, social-selling-and-dm, collabs-and-cross-promotion, ugc-and-influencer, creator-monetization, engagement-routine, community-management, crisis-and-moderation, scheduling-and-queue, platform-specs-and-validation, analytics-and-reporting, content-audit, experimentation-and-ab-testing.
Design decisions derived from the research
Section titled “Design decisions derived from the research”| Reusable lesson | Implementation in ReachWeaver | Verification state |
|---|---|---|
| Progressive skill specialization | One slim router SKILL.md with 5 on-demand references and three starter assets | Implemented, skill validator pending |
| Identity-first content | Voice/brand card template and separation rules | Implemented as guidance; no profile auto-extraction yet |
| Verified creative, native repurposing | Evidence and format protocols, conservative lint for 12 platform draft families | Implemented as guidance + basic lint; qualitative output still requires editor judgment |
| Actual publishing with receipts | Text-only adapters for Bluesky, Mastodon, Telegram (Update 2026-10-09: and LinkedIn member text, plus an optional Postiz backend); CLI stores remote IDs | Mock-tested; live validation pending, no real receipt |
| Avoid ghost-posts or duplicates | State machine, dry-run opt-in, atomic claim, needs_review and reconciliation | Unit-tested locally |
| Privacy and access control | Ignored JSON mapping, environment-only credentials, user login/MFA, safe API errors | Unit-tested partly; no OAuth vault |
| Attention learning | Latest-per-post native export snapshots; metric-specific coverage | Unit-tested; automated fetch pending |
| Crisis / kill switch | Pause/resume command and event log | Unit-tested |
| Keep provider choice open | Native adapter boundary and separate optional Postiz/MCP roadmap | Architecture defined; external adapters pending |
| Verify platform capability | Dated platform matrix and checklist per adapter | Documented; no 2026 API apps approved here |
Current platform restrictions relevant to architecture
Section titled “Current platform restrictions relevant to architecture”- Bluesky: official documentation describes
app.bsky.feed.postrecords and UTF-8 byte offsets for rich-text facets: https://docs.bsky.app/docs/tutorials/creating-a-post - Mastodon:
POST /api/v1/statusesaccepts bearer tokens andIdempotency-Key(stored for up to one hour): https://docs.joinmastodon.org/methods/statuses/ - LinkedIn: versioned Posts API with account-specific scopes, content-type distinctions and required headers; implement only after receiving app approval: https://learn.microsoft.com/en-us/linkedin/marketing/community-management/shares/posts-api
- TikTok: Direct Post requires approved
video.publishscope; unaudited clients restrict posts to private visibility: https://developers.tiktok.com/docs/en/content-posting-api-get-started - YouTube: videos uploaded by unverified API projects created after 2020-07-28 are private until compliance audit: https://developers.google.com/youtube/v3/docs/videos/insert
- Claude Code: project and personal skills and invocation model: https://code.claude.com/docs/en/skills
Decision on licenses and aggregation
Section titled “Decision on licenses and aggregation”Do not vendor all 106 skill directories or install third-party publishing scripts merely because they are public. Even permissive source has attribution and maintenance costs; some components have differing license obligations. The approach used here is conceptual synthesis + original code + citations + minimal external runtime dependencies. Evaluate each future imported module for license, correctness, privilege footprint and security against its actual upstream commit before inclusion.
Follow-up evaluation program
Section titled “Follow-up evaluation program”- Challenge with real user prompts (new release, multilingual paper, account recovery, platform-specific format, analytics anomalies).
- Test hallucination/no-source/no-permission cases and scripted action limits.
- Add agent evals of high/medium/low quality posts against a human-reviewed rubric; do not overfit to token heuristics.
- Perform scheduled quarterly checks of provider APIs, account-scoped permissions, social publishing rules, licensing and skill catalogue drift.
- Compare campaigns on real first-party metrics using matched post age/cohorts, rather than promising a universal best time/algorithm trick.
