openclaw-guardian-ultra deploys a bash watchdog for openclaw gateway with automated health checks, doctor-based repair attempts, git-backed workspace rollback, daily snapshots, and optional discord notifications.

eval-harness provides a structured framework for test-driven development in Claude Code sessions, covering eval definition, execution, and reliability metrics (pass@k) for regression and capability testing across model versions.

icloud-findmy queries find my device locations and battery status for family members via the pyicloud cli, requiring one-time apple id authentication and offering parsing patterns for coordinates and charge state.

starts work on marshroom cart issues by creating feature/hotfix branches, injecting issue context, and atomically updating status to running in the state file.

flutter-localization covers i18n/l10n setup for multi-language apps, including dependency injection, .arb file management, code generation via l10n.yaml, root widget configuration, and ios platform-specific locale registration.

feishu-img-tool wraps the two-step feishu image api (upload then send) with three actions: upload only, send with existing key, or combined upload-and-send. supports multiple formats up to 10mb with optional captions.

safe-config-momo enforces a structured workflow for modifying openclaw.json with mandatory backup, sanitization, validation, and explicit user confirmation via a specific passphrase before any changes are written.

moltmarkets-trader screens prediction markets, forms independent probability estimates, detects edge via threshold comparison, sizes positions using fractional kelly criterion, executes trades, and resolves outcomes. covers market creation, position tracking, and calibration logging.

skill-release-lifecycle provides a gated release workflow for clawhub skills, covering readiness checks, post-publish verification, and feedback classification. lightweight and operational but does not automate publishing or stability certification.

csharp-xunit covers test structure, data-driven testing with theory attributes, assertion patterns, and mocking strategies for building reliable unit tests in .net projects.

five ai agents compete in a daily evolutionary tournament on okx demo, trading live prices with exchange-level stop-losses. losers replaced by backtested strategies. runs 24/7 with manual evolution approval gate.

phone-controller automates android interactions via natural language, routing commands through adb and glm api. execution requires pre-confirmation for writes and avoids sensitive apps like banking.

debugging-dags provides structured root cause analysis for airflow dag failures using a four-step investigation workflow, failure categorization, and tool-specific commands for astro and oss deployments.

truenas-skill provides api-level control over truenas scale storage pools, datasets, snapshots, alerts, and containerized apps via rest and websocket endpoints. covers dockge orchestration and integrates references for homelab media and download services.

pr-reviewer analyzes github pull request diffs for security, error handling, and style issues across go, python, and javascript. tracks reviewed state and outputs markdown reports with categorized findings.

migpt-xiaomi-assistant covers deployment of a custom llm-powered voice assistant on xiaomi smart speakers, replacing the built-in mina ai. includes critical troubleshooting for tts failures, authentication loops, and wake word conflicts.

doc handles read, create, and edit workflows for docx files with optional visual layout validation via pdf conversion. python-docx powers structural edits while fallback text extraction mitigates missing rendering tools.

flow-crm wraps the flowdeck rest api for client, deal, proposal, receivable, expense, and contact operations. supports crud actions with structured filtering and mandatory pre-flight data collection for client creation.

flow-search wraps claude's native search capability for multi-source web research and synthesis, returning structured reports with sources. targets market analysis and competitor intelligence over quick lookups.

agentdrive-cli provides shell access to cloud storage with structured json output, command mapping tables, and cross-platform pipeline examples. covers directory browsing, file upload/download, search, sharing, and config manipulation.

solo-cli wraps the SOLO.ro accounting platform with command-line access to summaries, revenues, expenses, invoices, and e-factura documents. covers installation, configuration, and a dozen subcommands via both CLI and interactive TUI mode.

acestep-lyrics-transcription converts audio files to timestamped lyrics via openai whisper or elevenlabs scribe api, outputting lrc, srt, or json formats. includes mandatory post-transcription correction workflow to fix common transcription errors on sung lyrics.

robot-evolve executes low-risk autonomous maintenance (L0/L1 ops) on manual user trigger, cleaning temp files, validating workspace structure, and compressing memory logs when thresholds are exceeded. reports sent via current session channel.

trade-it wraps the Trade It API to execute stock, options, and crypto trades across eight major brokerages. it handles draft creation, user confirmation flows, and brokerage connection via hosted URLs.

confidential-agentic-payment-stack enables encrypted erc-7984 token transfers and escrow workflows for autonomous agents on ethereum sepolia/mainnet using zama fhEVM, with support for identity registration and delegation-based balance viewing.

agent-wallet-usdc enables ai agents to create multi-chain wallets, check balances, transfer usdc and native tokens, and bridge assets across solana, base, and ethereum. wallet derivation uses a single bip-39 seed phrase stored in environment variables.

web-scraping covers cascade fallback strategies, poison pill detection, and ethical extraction patterns for javascript-heavy sites, undocumented apis, and platform-specific scrapers like youtube and instagram.

memclawz-connect exposes a shared memory bus for multi-agent fleets via http api, supporting read-before-act and write-after-complete patterns across nine memory types. agents search for context, then record outcomes to a centralized qdrant store.

byted-bp-cdn-pagesdeploy automates static site deployment to BytePlus Edge Pages with one-command create-or-update, custom domain binding, and CDN acceleration. requires AK/SK configuration and index.html presence.

lsp28-grid covers creating and updating grid layouts on lukso universal profiles with support for mini-apps, iframes, and external links. encodes grid data as verifiable uris and executes via keymanager contracts.

claw-wallet-pro provides multi-chain sandbox signing with local PIN control and policy-driven asset transfers. requires explicit user confirmation for transactions and bootstraps via shell scripts. covers wallet binding, balance inspection, and defi interactions through a local http api.

an-elegant-defense maps matt richtel's immune system framework into five executable use cases: immune basics, autoimmune disease, chronic inflammation, immunotherapy, and immune aging. covers core concepts (t cells, b cells, antibodies) and lifestyle interventions with clear routing rules but lacks step-by-step procedures.

agmsg provides a messaging and discovery layer for autonomous agents on base blockchain, supporting private chats, groups, channels, and search with per-action x402 micropayment settlement. agent-only network with cli tooling.

openclaw-comfyui-api-runner connects to comfyui servers via saved or ad-hoc profiles, accepts workflow json, submits jobs, polls for completion, downloads outputs, and serves them locally with view urls.

cn-model-gateway wraps ten chinese llm providers (deepseek, qwen, glm, kimi, hunyuan, doubao, minimax, 01, baichuan, stepfun) behind a single mcp server exposing ask_model, compare_models, list_providers, health_check tools plus usage stats and prompt templates. zero external dependencies, hardware-aware concurrency, unified error mapping to chinese messages.

crewai covers multi-agent orchestration with role-based personas, task dependencies, and sequential or hierarchical coordination. includes yaml config patterns, planning features, and anti-pattern guidance for agent design.

repo-runner clones github projects and bootstraps them by extracting prerequisites from documentation, detecting project type, and running appropriate install and execution commands with user confirmation gates on untrusted operations.

costhq tracks agent session costs, enforces budget limits, and logs ai usage across cloud and local models. provides a web dashboard, semantic caching proxy, and soc2-compliant audit trails for enterprise deployments.

compress applies iterative semantic compression with anchor validation and multi-level reliability tiers. production use limited to l1-l2 to avoid information loss on critical content.

adversarial-spec transforms brief inputs into formal specifications via a two-role adversarial pipeline that writes, challenges, and revises specs with yaml frontmatter and acceptance criteria. git-aware with squash-merge on approval.

jina-reader extracts clean web content in three modes (read, search, fact-check) via the jina ai api, routing requests through their infrastructure to avoid exposing server ip.

dnote provides cli-driven note capture, search, and organization with optional cloud sync. covers add/retrieve/edit/delete operations across named books, useful for building searchable reference libraries during development.

whatsapp-private-domain-ops provides a compliance-first strategy framework for whatsapp account health diagnostics, tiered user management, and three-layer adaptive operations planning within meta's policy boundaries. purely advisory, no api execution.

nudge-cli covers installation, core commands (task/secret/punishment), and structured onboarding flow for setting up an accountability tool with social consequence mechanics.

watermark-remover-skill automates a local PyQt GUI program by copying images to a monitored input folder and polling for results in the output folder. requires gui to be running with directory monitoring enabled and valid config paths.

second-brain-ai captures, searches, and assembles context from local markdown vaults with controlled writes and attribution. file-scan based, no semantic search or automatic indexing.

baidu-drive provides file management operations (upload, download, transfer, share, search, move, copy, rename, mkdir) scoped to /apps/bdpan/ plus agent memory backup/restore for claw environments. execution requires pre-flight checks (install, login, path validation) and explicit user confirmation for high-risk operations.