Draft: Friday Reel — MP4 ready to queue on your Facebook Page.
OpenQuok Brand Page @ 9:00 AM (in 4 days)
Note: Review agent caption before scheduling
OpenQuok ThinkRail
Schedule social media from ThinkRail then you approve
ThinkRail hosts the pi coding agent in a real IDE — git worktrees, Monaco tabs, terminals, and concurrent chats. Add the openquok-core skill so the agent drafts and schedules social posts while you review and approve on the calendar or kanban.
Install our core skill:
mkdir -p ~/.pi/agent/skills/openquok-core
curl -fsSL "https://raw.githubusercontent.com/Ratimon/openquok-monorepo/main/agent/skills/openquok-core/SKILL.md" \
-o ~/.pi/agent/skills/openquok-core/SKILL.mdDraft and schedule my product launch thread for Tuesday at 9am across Facebook, Instagram, and
Threads.
On it — openquok-core
Created drafts with cover image attached.
- • Facebook — Tue 9:00 AM
- • Instagram — Tue 9:00 AM
- • Threads — Tue 9:00 AM
✓ Queued for Tue 9:00 AM · ref oq-7f3a
Your worktree IDE
Chat with pi in ThinkRail,keep main clean
Open a repo, cut a workspace worktree, and ask the agent to draft and schedule like any other task. The openquok-core skill runs in that worktree’s shell, finds connected channels, attaches media, and queues drafts — you approve on the calendar before anything publishes.
Built for ThinkRail users
Who connects ThinkRail to OpenQuok?
Worktree-first operators
Each workspace is a real git worktree — own branch, files, terminals, and chats. Install the CLI once, drop openquok-core into ~/.pi or .pi/skills, and schedule without mixing drafts into main.
Pi coding agent users
ThinkRail is a thin host: pi owns models, skills, and cost. OpenQuok stays CLI-first so you reuse the same recipes in ThinkRail, a pi CLI session, or another skill host.
Founders & small teams
Keep human approval on the calendar while the agent handles volume from a dedicated worktree. Specs stay beside the code; social drafts stay behind OpenQuok review.
How it works
Five steps,to ThinkRail + OpenQuok
Connect once
authenticate in the worktreepick your workspacechat from the IDE securely
Choose an OpenQuok workspace, connect with OAuth2 — approve in your browser, and credentials stay on the machine that runs ThinkRail. Ask the agent in chat to draft and schedule without opening another app.
CLI authentication options
# OAuth2 device flow (interactive — opens browser)
openquok auth:login
openquok auth:statusOn-going Tasks
Review agent and human drafts across Facebook, Threads, Instagram, and more — drag posts between Drafted and Scheduled.
Drafted posts
Draft: Carousel draft — four images queued for Instagram feed.
OpenQuok Brand @ 2:00 PM (in 8 days)
Note: Confirm cover image
Draft: Vertical video draft — trending audio added in TikTok app.
OpenQuok @ 6:00 PM (in 12 days)
Note: Pick sound in TikTok before publish
Draft: Launch tweet with follow-up reply chain — agent draft.
OpenQuok @ 3:30 PM (in 16 days)
Note: Review reply timing
Scheduled posts
Drop here to schedule
Published posts
Published posts stay in this column
Kanban + smart filters
Review every AI draftsign off confidentlybefore it goes live
Chat, move agent-generated posts from draft to review to scheduled on a kanban board—with the same smart filters as your calendar. Approve quality at scale instead of trusting autopilot.
CLI command options
# Draft + human checklist
openquok posts:create -c "…" -s "…" -t draft -i "<uuid>" --note "Check CTA before schedule"
openquok posts:review-todo <post-id> --note "…"
openquok posts:status <post-id> --status draft
openquok posts:status <post-id> -s scheduleAnalytics
Ask what workedsee winnersand adapt from chat
Ask the agent in ThinkRail to pull impressions, likes, comments, and shares for any connected channel. Compare performance and schedule more of what already resonates — without opening the dashboard.
CLI analytics options
# Platform metrics (followers, impressions, engagement)
openquok analytics:platform <integration-uuid> -d 30
# Per-post insights (likes, comments, shares)
openquok analytics:post <post-id> -d 7What performed best on my channels over the last 7 days? Break down impressions and engagement.
Pulled platform analytics (7d)
Top post: 12.4k impressions, 842 reactions — carousel on product tips.
Facebook Page leads with +24% impressions vs last week.
Compare that post on TikTok vs Instagram Reels.
TikTok: 48k views, 2.1k likes. IG Reel: 31k plays, 312 saves. TikTok wins reach; IG leads on saves.
Draft and schedule my product launch thread for Tuesday at 9am across Facebook, Instagram, and
Threads.
On it — openquok-core
Created drafts with cover image attached.
- • Facebook — Tue 9:00 AM
- • Instagram — Tue 9:00 AM
- • Threads — Tue 9:00 AM
✓ Queued for Tue 9:00 AM · ref oq-7f3a
What performed best on my channels over the last 7 days? Break down impressions and engagement.
Pulled platform analytics (7d)
Top post: 12.4k impressions, 842 reactions — carousel on product tips.
Facebook Page leads with +24% impressions vs last week.
Compare that post on TikTok vs Instagram Reels.
TikTok: 48k views, 2.1k likes. IG Reel: 31k plays, 312 saves. TikTok wins reach; IG leads on saves.
Draft and schedule my product launch thread for Tuesday at 9am across Facebook, Instagram, and
Threads.
On it — openquok-core
Created drafts with cover image attached.
- • Facebook — Tue 9:00 AM
- • Instagram — Tue 9:00 AM
- • Threads — Tue 9:00 AM
✓ Queued for Tue 9:00 AM · ref oq-7f3a
Scale what works
when a format hits, scale by adding worktrees and parallel chats
Spot a winner in analytics, then cut another worktree for the next client or brand while a second pi session queues the next wave — credentials, channels, and agent context stay isolated as you scale.
Parallel CLI sessions
# Workspace A — launch (client brand)
openquok posts:create -c "…" -s "…" -t draft -i "<uuid>"
openquok posts:status <post-id> -s schedule
# Workspace B — another client (isolated credentials)
openquok posts:list --status draft
# Same workspace — metrics in parallel
openquok analytics:platform <integration-uuid> -d 7
openquok analytics:post <post-id> -d 30Playbooks & Building Blocks Directories
Explore Viral Formats
Mix playbooks, skills, and MCP servers—then tailor them to your use case.
Playbooks
ChatGPT Video Edit & Publish
Brief in ChatGPT, edit video with CapCut CLI or Sora, then schedule via OpenQuok MCP — you approve before anything go…
Viral TikTok Carousel
Research → lock character → generate → overlay → post via openquok-core. Install openquok-core + openquok-tiktok-slid…
Skill Builder
Build and export a SKILL.md from building blocks.
See All
Browse every published playbook.
Building Blocks
Capcut Edit
Edit CapCut / JianYing video projects — read and write subtitles, timing, speed, volume, templates, animations (fade/…
OpenQuok TikTok Slideshow
Research a channel, lock a character, generate portrait slides, overlay text, and post via openquok-core. Install wit…
OpenQuok Core
Schedule and manage social posts with the openquok CLI — authenticate, upload media, create drafts and scheduled post…
Bloom MCP
Onboard your brand once, then generate and edit on-brand images from any MCP client — website or Instagram URL in, ca…
RevenueCat MCP
Wire subscription metrics and entitlement state into your agent — products, offerings, and customer data over HTTP MC…
See All
Browse every published building block.
comparisons
agent-native scheduling, not another dashboard
Most social scheduler SaaS keeps you in a browser tab. OpenQuok is built for agents
Typical social scheduler SaaS
- Copy posts between your AI chat and a separate scheduling tool
- Siloed API keys and workflows that do not compose with your agent stack
- Always-on integrations that bloat agent context
- Locked to one vendor's models or automation layer
- One working tree mixing every experiment as you add channels and parallel sessions
- Autopilot publishing with no human checkpoint
OpenQuok + ThinkRail
- Ask the agent in ThinkRail to draft and schedule from the worktree
- Credentials stay on the machine that runs ThinkRail — workspace isolated
- Skills load on demand from ~/.pi or the worktree .pi/skills folder
- CLI-first openquok-core works beside any other pi skill or extension
- Git worktrees isolate each line of work — merge the good branch, delete the rest
- Every post lands as draft or scheduled — you approve before anything goes live
CLI
Command reference
Commands from the openquok-core skill — structured JSON on stdout, human-in-the-loop drafts, and workspace media uploads.
openquok auth:login --jsonStart device OAuth for agents; user then opens verification url in browser
openquok integrations:listList connected social media accounts
openquok integrations:settings <integration-id>Get platform-specific settings schema and posting rules
openquok upload <file>Upload media and get a reusable id and path
openquok posts:create -c "…" -s "…" -i "<integration-id>"Create and schedule a new post
openquok posts:listList scheduled and draft posts
openquok posts:status <post-id> -s draftMove a post between draft and scheduled before publish
openquok analytics:platform <integration-id> -d 30Get analytics for an integration or channel
openquok analytics:post <post-id> -d 7Get analytics for a specific published post
openquok plugs:catalogList global plug types per provider (likes-threshold rules)
openquok plugs:list <integration-id>List saved global plug rules on a connected channel
openquok plugs:upsert <integration-id> --func autoPlugPost --fields '[{"name":"likesAmount","value":"100"},{"name":"post","value":"Thanks!"}]'Create or update a channel-level plug rule (auto-reply when likes hit a threshold)
openquok plugs:activate <plug-id> --activated falseEnable or pause a global plug rule
Where you work
Worktrees, editor, and terminals
ThinkRail scopes files, diffs, terminals, and pi chats to the active git worktree — then invoke openquok-core like any other skill:
macOS
CLI or desktop build on Apple Silicon
Linux
CLI or desktop on arm64 and x64
Windows
CLI or desktop build for Windows x64
Git worktrees
Each workspace is its own branch and cwd
Monaco editor
Tabs, files tree, and live diffs per worktree
Worktree terminals
PTYs scoped to the active workspace cwd
Optional MCP extensions
Pi MCP extensions
Optional MCP tools beside skills — OpenQuok stays CLI-first
By channel
ThinkRail by channel
Choose a channel for platform-specific workflows, examples, and FAQs.
Soon
Page feed posts, MP4 Reels, and link-preview cards — built for Facebook Pages, not personal profiles.
Threads
Soon
500-character posts with media and scheduled follow-up reply chains — conversation-first text on Meta Threads.
Soon
Feed posts, Reels, carousels, and Stories — with trial Reels, collaborator tags, and follow-up comments.
YouTube
Long-form MP4 uploads and vertical Shorts — title, privacy, tags, and custom thumbnail per video.
TikTok
Vertical video and photo carousels — direct publish or queue to your TikTok inbox to add trending audio.
B2B thought leadership on profile and company Page — scheduled, reviewed, and ready when buyers research you.
X
Weighted 280-character tweets (4000 for Verified), up to four images or one video, and scheduled thread replies.
Dev.to
Schedule Dev.to posts with tags and series — plus cover, organization, canonical URL, and analytics for page views, reactions, and comments.
Frequently asked questions
ThinkRail + OpenQuok,answered
What ThinkRail is, how it relates to the pi coding agent, how to install openquok-core, worktree credentials, human approval, and why CLI beats MCP-only for scheduling.
ThinkRail is a worktree IDE for the pi coding agent. You open a git repo as a project, cut workspaces as git worktrees, and work across Monaco tabs, terminals, diffs, and concurrent pi chat sessions — all scoped to the active worktree.
ThinkRail is a thin host: pi runs in-process and owns models, skills, compaction, and cost. ThinkRail owns the workspace, the editor, and the UI. You do not need a separate /agents/pi landing — install openquok-core where pi discovers skills, then chat from ThinkRail. A standalone pi CLI session can use the same skill files.
In a worktree terminal, run npm install -g @openquok/auto-cli@latest, then curl SKILL.md into ~/.pi/agent/skills/openquok-core/SKILL.md (all projects) or .pi/skills/openquok-core/SKILL.md in the worktree. Start a new chat so pi reloads skills. Set OPENQUOK_API_KEY from Account → Settings → Developers → Access or run openquok auth:login. See the ThinkRail agent guide or the scheduling walkthrough for exact commands.
The global CLI and auth files live on the machine that runs ThinkRail — the same environment as worktree terminals. Your OpenQuok workspace tokens never need to be pasted into chat; use OAuth device flow or a programmatic opo_ token on that computer.
The agent can draft and schedule posts, upload images and video, apply per-platform settings, schedule threads and follow-up comments, configure internal and global plugs, and pull platform and post analytics — across every channel connected in your workspace. The openquok-core skill documents integrations:list, posts:create, plugs:upsert, posts:status, analytics:platform, upload, and more; every command returns structured JSON for the agent to parse.
YouTube, TikTok, LinkedIn, and X are available today. Facebook, Instagram, and Threads are coming soon. Connect channels in the OpenQuok web app; ThinkRail uses integration UUIDs from openquok integrations:list to target the right accounts.
Posts created through the CLI land in your OpenQuok workspace as drafts or scheduled items — nothing goes live on autopilot. You review on the calendar or kanban, move posts through draft and review, and approve what should publish. ThinkRail handles volume; you handle quality.
ThinkRail is not a native OpenQuok MCP client like Cursor. Pi loads skills and can use optional MCP extensions, but openquok-core is CLI-first: structured JSON, media uploads, plugs, and analytics in one skill. MCP fits point-in-time tool calls in editors that host OpenQuok MCP; the CLI skill fits repeatable scheduling recipes in a worktree shell. For editor-native MCP scheduling, see OpenQuok for Cursor. Many teams use both — MCP for ad hoc tools in another IDE, openquok-core inside ThinkRail.
OpenQuok offers a 7-day free trial for scheduling. ThinkRail is open source; you bring your own pi provider credentials. Install ThinkRail, add openquok-core, and begin scheduling from a worktree chat.
Read the ThinkRail setup guide
Step-by-step instructions for connecting OpenQuok to ThinkRail — install commands, auth patterns, and configuration snippets.
Start free. Publish with confidence.
Connect your agent, review every draft, and schedule posts across channels before anything goes live.