Supernor/openclaw-skills - Page 2

log-diagnostics

Scan gateway logs for recurring error patterns, group and classify findings, escalate critical issues to Reactor.

morning-briefing

Own and improve Robert's morning briefing. Pull from /api/digest (zero tokens), format compact, link to Bridge. Learn what Robert engages with vs ignores.

ai-news

Produce daily AI news digest for the team

research-before-build

Research methods, libraries, and patterns BEFORE coding. Prevents building the wrong thing.

research-estimate

Estimate cost of a research task before executing it. Gate expensive tasks with human approval.

research

Deep research on a topic using web search and analysis capabilities

research-troubleshoot

Diagnose why a task or agent failed. Check local data first (zero tokens), web search only if needed.

research-upstream

Track OpenClaw upstream changes, releases, and leverage opportunities. Weekly audit.

web-search

Queue a web search via Gemini CLI on the host. Free tier, no API cost. Results return via ops.db.

discord-project

Create a new Discord project channel in the Projects category, initialize decision tracking, and add to gateway allowlist. Usage: /project <name>

build-architect

Design project architecture from specs. Reads project.md, writes architecture.md.

build-builder

Build project files from architecture. Reads architecture.md + project.md, creates all files.

build-deployer

Detect project type, deploy it, return test URL accessible from any device.

build-reviewer

Review built project files for quality. Flags critical issues only.

data-path-audit

Audit task routing quality, engine success rates, and token spend. Weekly duty for Ops Officer.

discord-components

Send Discord messages with native buttons, select menus, modals, and polls using OpenClaw component format.

gauntlet-orchestrate

Run the 3-agent Gauntlet debate. Codex (devils advocate) vs Reactor (positive advocate), Scribe mediates. Two rounds, then user decisions.

model-api-speed-test

Benchmark model APIs for speed, quality, and stream stability. Configurable for any endpoint.

project

APS project management — create, view, list, validate, delete, undelete, history. Template-driven, all output JSON.

tool-audit

Audit and update all agent TOOLS.md files for MCP tool awareness. Self-improving — learns from each run.

workshop-dispatch

Split a Green Light idea into properly-sized automation tasks with dependencies. One task = one thing.

workshop-monitor

Track Workshop ideas through stages. Flag stalled ideas, report pipeline health, suggest next actions.

idea

Guided idea intake via Telegram or Discord. Button-first, typed-fallback. Per-user voice.

import-conversations

Import Corinne's ChatGPT conversation history into Eoin's memory. Accepts pasted text, Google Docs links, MD files, or share links. Reflects back what was learned.

onboard

First-impression onboarding conversation for Corinne. Guides her through personality setup, goal capture, and ChatGPT import in a natural, conversational flow.

prepare

Preparation check — am I ready for what Corinne will do next? Reviews memory, prefs, onboarding status, and known gaps. Reports readiness and flags what needs attention.

video-discuss

YouTube video analysis and discussion for Corinne. Ingests video, gets leverage analysis from Strategist, presents findings in plain language, drives conversation about ideas.

vision-capture

PTV interview — multi-turn conversation where Eoin interviews Corinne about family goals and outputs Purpose Toward Vision codes. Sends to Robert for async validation via bearings.

video-discuss

YouTube URL handler. Step 1: call exec tool with command "video-discuss <url>". Step 2: respond with EXACTLY 3 lines, nothing more: Line 1: **{title}** by {channel}. Line 2: One single sentence summarizing the summary field. Maximum 25 words. One period.…

video-analyze

OpenClaw-aware leverage analysis of video transcripts. Produces structured JSON with leverage points, easy wins, cost analysis, discussion seeds, and auto-proposed ideas.

telegram-transcript

View Telegram conversation history and search past messages for truth verification

context-capture

Capture a structured context snapshot for the Black Box system. Gathers recent charts, agent sessions, health events, ops activity, and taint status into a single snapshot charted for future recovery.

chart-batch

Batch verify, update, and archive stale charts with consistency checks and clear status output. Use for large chart maintenance passes.

diff-report

Produce a delta report since a timestamp, covering new charts, config changes, and agent activity. Use when Reactor asks what changed since a prior checkpoint.

helm-health

Report on Helm routing health — escalation rate, learned patterns, routing efficiency. Use when asked about engine routing performance or during sitrep generation.

preflight

Build pre-session context showing what changed since last session, including new charts, alerts, and notable agent/system activity. Use at session start before substantive work.

script-gen

Generate Python or bash scripts from natural-language specs with practical defaults, safety checks, and usage notes. Use for repeatable automation and grunt-work tasks.

sitrep

Generate a concise system situation report (health, agent status, stale issues, pending ideas, transcript signals) and write it to /root/.openclaw/sitrep.md. Use when Reactor needs fast current-state context.

validate

Pre-validate config or code changes using syntax checks and dry-runs before risky operations. Use immediately before apply/restart/deploy actions.

check-in

Alignment check-in with Robert via Discord. Captures intent notes, vision/PTV direction, and functionality feedback. Robert's equivalent of Eoin's onboard skill.