socai-cli
Skill256 starsChanged 46 days agoApache-2.0View on GitHub
Use the installed socai CLI for research on Xiaohongshu, Douyin, TikTok, Instagram, LinkedIn, and podcasts. Use when an external agent needs real posts, comments, creator profiles, or source-linked social research through the user's signed-in browser, or podcast summaries and transcripts without a browser.
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- clean-targetsSkill
- socai-model-syncSkill
- socai-releaseSkill
- socai-site-deploymentSkill
- rsiSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use socai --help and socai <site> <command> --help as the authoritative command reference.
- Very specific: line 18
Before the first site command for a new user request, pass the user's original question to socai task begin.
- Very specific: line 47
If the original question is no longer available, use user_prompt: null rather than reconstructing it.
- Distinctive: line 61
Treat login, permission prompts, and platform verification as access limitations.
- Distinctive: line 65
Keep research read-only unless the user explicitly requested a specific write.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
socai CLI research
Use socai --help and socai <site> <command> --help as the authoritative
command reference. socai status --json checks readiness without connecting
Chrome. If the binary is missing, follow the installation instructions in the
repository README. Do not substitute scraping or private APIs for socai.
Start each new task once
This is the standard workflow for all platform CLI operations. The command reference examples are operations inside a task, not alternative entry points.
Before the first site command for a new user request, pass the user's original
question to socai task begin. Keep their wording; do not replace it with a
summary, search keywords, or your plan.
socai task begin "Find why first-time campers regret their gear purchases."
socai xhs search "camping gear regrets" --num-notes 10 > /tmp/camping-xhs.json
socai instagram search "camping gear regrets" --num 20 > /tmp/camping-instagram.jsonOrdinary commands automatically join the daemon's current task. No ID or extra
argument is needed. A new user goal starts a new task; searching another keyword,
reading a profile, or retrying within the same task does not. Register again
after the daemon restarts. Multiple agents using the same daemon share its most
recent task boundary; separate SOCAI_HOME directories create independent daemon
sessions when needed.
For multiline or shell-sensitive text, write a task-specific UTF-8 JSON file using
a file tool or JSON serializer, then run socai task begin --context-file <path>:
{
"user_prompt": "The user's original question",
"agent_host": "unknown"
}agent_host is optional; use the actual host identifier when known. Stdin is
supported with --context-file -. Remove the temporary input file after use.
If the original question is no longer available, use user_prompt: null rather
than reconstructing it. Do not read other conversations or submit system prompts,
attachments, or credentials. Registration does not connect Chrome.
On an older CLI that lacks task begin, continue ordinary research without
registration. Do not repeatedly retry or automatically upgrade. Failed
registration leaves the previous daemon task unchanged.
Gather and deliver evidence
Choose among xhs, dy, tiktok, instagram, linkedin, and podcast according to the
user's scope. Inspect command help before using platform-specific options.
Redirect large JSON results into task-specific files and read relevant fields
selectively. Keep source URLs and distinguish collected evidence from inference.
Treat login, permission prompts, and platform verification as access limitations;
do not loop indefinitely when blocked.
Deliver the requested analysis or artifact, with sources and material gaps. Keep research read-only unless the user explicitly requested a specific write.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

