@austindixson

better-ralph executes one prd-driven coding iteration: reads story priority, implements acceptance criteria, runs quality checks, commits, updates prd.json and progress.txt. uses only standard tools (read, write, exec, git). requires prd.json in workspace root.

dacker installs docker reliably across macos, linux, and windows, then enables building, running, and composing containers via step-by-step procedures grounded in official docs.

skill-safety-checker evaluates downloaded skills against a structured 8-category security framework (purpose alignment, registry consistency, RCE patterns, malicious code, credentials, persistence) and produces a benign/suspicious verdict with actionable remediation guidance.

docs-bot fetches current documentation for api integrations and coding tasks, prioritizing live sources over training data. covers openrouter in detail, with fallback patterns for other services.

agent-loops chains multi-agent workflows for software development tasks (build, debug, review, research, refactor). outputs flow between sequential and parallel steps with role-specific prompts and model routing.

visual-explainer-main generates self-contained html pages for technical diagrams, tables, and visualizations. covers architecture, flowcharts, schemas, and data tables with guidance on aesthetic choices, mermaid theming, responsive patterns, and optional ai image generation.

create-agent automates registration of new overstory agents across seven integration points (config, manifest, definitions, gateway prompt, task router, context generation, and regeneration). supports manual creation and optional analysis from logs and project structure.

project-manager-agent monitors openclaw sub-agents for stalls, detects failures, and can steer stuck sessions via sessions_send. requires agent to parse json output and manually invoke steering calls.

overstory-integration bridges claude code agents with nanobot orchestrator via lifecycle tracking, sqlite mail routing, and git worktree coordination. covers agent spawning, status monitoring, inter-agent messaging.

skill-doc-formatter normalizes SKILL.md files into a consistent structure (Description, Installation, Usage, Examples, Commands) for consistent rendering on ClawHub. includes optional security review checks for common issues flagged by ClawHub scans.

launchagent-manager lists, classifies and prunes macos launch agents, ensuring the openclaw gateway plist stays loaded with matching tokens. supports dry-run and apply modes for safe cleanup of non-openclaw daemons.

what-just-happened reads gateway logs after restart and posts a brief summary to tui or telegram. it detects restart type (clean restart, sigusr1, auth change, reconnect) and optionally recommends gateway-guard for remediation.

nanobot-overstory-bridge routes tasks from an ollama mistral orchestrator to a claude code agent swarm via http gateway, maintaining session mappings and syncing memory bidirectionally between systems.

auto-clipper monitors a folder for media files, delegates clip analysis to agent swarm, and generates trimmed segments via ffmpeg. supports cron scheduling and batch processing with configurable output organization.

scribe generates daily and weekly markdown summaries by scanning openclaw logs, chat history, memory files, drafts, and behavior configs. output includes structured summaries and detailed activity breakdowns.

subagent-tracker surfaces progress and status of running subagent swarms via voice commands like 'agent status'. provides list, status, and tail transcript operations for visibility into queued tasks and active sessions.

chat-history-analyzer extracts cursor ide conversation logs, identifies key discoveries/obstacles/solutions via pattern matching, and writes markdown reports to a journal directory. designed for hourly cron execution.

skills-compat loads and registers tool definitions from openClaw skill.md files, mapping them to nanobot, overstory, and ollama formats via a singleton registry. architecture is sound but implementation details are absent.

blog-generator scans journal entries to identify high-value topics and produces blog posts in html format. it includes topic scoring heuristics and optional humanization via external api, but lacks explicit failure handling and actionable input contracts.

creative-agents wraps four agent types (researcher, social media, blogger, scribe) with python integration scripts for skill discovery, oauth, browser automation, and log analysis. lacks concrete execution procedures and error handling.

subagent-dashboard provides a flask-based web ui for monitoring openclaw subagent sessions, transcripts, and task progress with 3-second auto-refresh and stalled-agent detection over 30 minutes.

playwright-commander wraps browser control via playwright python bindings, offering granular dom interaction, screenshot capture, and javascript execution for ui automation tasks. incomplete specification and vague error handling.

doppleganger wraps subagent-tracker's duplicate detection to prevent concurrent runs of the same task. returns exit codes and json payloads indicating task state, delegating actual deduplication logic elsewhere.

composio-composer-xskill wraps twitter/x posting via composio's http interface using beautifulsoup. covers tweet CRUD operations with token-based auth, but lacks error handling depth and concrete failure recovery.

self-optimizer scans openclaw logs and configuration to surface performance and stability recommendations, though core features remain unimplemented and procedures lack rigor.

tweet-crafter generates tweets and blog posts for openclaw projects with clawhub link enforcement, but lacks actionable procedures and handles only the happy path.

facepalm correlates openclaw console logs with recent chat history and applies codex 5.3 analysis to diagnose issues. triggered automatically on troubleshooting loops or manually via cli.

claw-canvas wraps openclaw's native canvas tool to display agent thinking, intermediate results, and rich content during development. the skill lacks concrete trigger definitions and actionable procedures.

Monitors OpenClaw cron job health, identifies failures, timeouts, and delivery issues.

Austin's intelligent model router with fixed scoring, his preferred models, and OpenClaw integration

Ensures OpenClaw gateway auth consistency. Use when checking or fixing gateway token/password mismatch, device_token_mismatch errors, or before delegating to...

IMPORTANT: OpenRouter is required. Routes tasks to the right model and always delegates work through sessions_spawn.

Installs and uses Docker reliably with official docs. Use when installing Docker (Desktop or Engine), building or running containers, writing Dockerfiles, us...

Mulch Self Improver — Let your agents grow 🌱. Captures learnings with Mulch so expertise compounds across sessions. Use when: command/tool fails, user corre...