We publish what we build. These are the skills and contributions we've shipped publicly — live on ClawHub and in the OpenClaw upstream.
ClawHub Skills8 skills published on ClawHub
Strips AI-generated patterns from text — em dashes, hollow openers, sycophantic phrases — and rewrites in a natural human voice.
View on ClawHub →Humanizes X/Twitter threads for authentic engagement — removes corporate tone, adds personality and rhythm.
View on ClawHub →Publishes content to social platforms via Buffer’s GraphQL API — X, LinkedIn, scheduling and thread support.
View on ClawHub →Web-backed fact-check and attribution audit for social content — verifies claims before publishing.
View on ClawHub →Hypothesis-driven AI model evaluation framework — structured experiments, metrics tracking, and result documentation.
View on ClawHub →Checks factual claims in text using web search and structured source citation.
View on ClawHub →IMAP bulk email triage — pattern-based delete/archive with dry-run mode.
View on ClawHub →Handles WebAuthn, email-link verification, and session persistence flows for new service signups.
View on ClawHub →We proposed Playbooks as a native OpenClaw feature: reusable, version-controlled specs for multi-agent pipelines — the equivalent of GitHub Actions reusable workflows for AI agents. The RFC was submitted upstream to the OpenClaw GitHub repository and discussed in the OpenClaw community Discord.
We also use playbooks internally — Build Pipeline, Insight-to-Social, Product Launch, Newsletter Intelligence, and more. The concept ships here before it ships anywhere else.
Documented standards from production experience — designed to be adopted, not just read.
A published specification for responsible spend governance in multi-agent teams. Daily token limits per agent, automatic midnight resets, yellow/red alert thresholds, and demotion protocols that reduce mesh privileges when budget limits are breached.
Built from production experience running 13 agents daily. Documented and open for adoption.
A two-layer session continuity spec for AI agents — STATUS.md per project (human-readable, fast) plus a structured task plane (persistent, queryable). Solves the cold-start problem: every agent reads the STATUS.md at session start. The file is truth; in-memory context isn't.
In production across all 13 lab agents. Documented and open for adoption.