clode-labs
- 46 skills
- 0 followers
- 1 week ago last updated
- ▌ Local Deployment · clode-labsFull local deployment protocol — expose ALL HTTP services (frontend + backends) via aramb expose FIRST to collect public URLs, inject those URLs as runtime env var overrides into docker compose (no file edits), then start the stack. Deliverable: verified public URLs for every exposed service, with inter-service URLs wired so the browser-side frontend hits the exposed backend. Use when: deploying an application locally for preview or end-to-end validation. NOT for: production deployments, code development, or testing.
- ▌ Configure Trigger · clode-labsConfigure an EVENT trigger on a workflow from a natural-language phrase — make a workflow fire when something happens in a connected service. Use when the user says things like "fire this when a new GitHub issue is created", "run this whenever I get a Slack DM", "trigger on every push to main", "stop firing on new issues", "remove the trigger". NOT for: wall-clock / cron schedules ("run this every Monday at 9am" → use schedule-workflow), creating or editing the workflow definition (use create-workflow / update-workflow), or triggering a single one-off run.
- ▌ Frontend Critique · clode-labsQA skill for frontend work. Validate implementation against requirements, run/test the feature, and trigger rebuilds when issues found. Use this skill to review UI components, test functionality, verify accessibility, and ensure quality.
- ▌ Schedule Workflow · clode-labsConfigure a cron schedule on a workflow from a natural-language phrase. Use when the user says things like "run this weekly", "every Monday at 9am", "stop the schedule", "pause it", "disable the schedule". NOT for: event-driven triggers ("fire when a new GitHub issue is created" → use configure-trigger), triggering a single one-off run, or editing the workflow definition.
- ▌ Backend Deployment · clode-labs bundleDeploy backend services using aramb.toml. Deploys TOML to create services, then builds and deploys backend images using aramb build with --service flag. Returns PUBLIC_URL for frontend integration. Frontend deployment handled by frontend-deployment skill.
- ▌ Frontend Detection · clode-labsDetect which service in a docker-compose.yml is the frontend. Use when: (1) need to identify the UI service for tunneling, (2) need to find the frontend port for testing, (3) determining which service serves the user-facing application. NOT for: backend service detection or general docker-compose analysis.
- ▌ Analyse Conversation · clode-labsRead and analyse an agent's real end-user conversations (aramb_agents. conversation_search / conversation_get) to judge how the persona actually behaves and turn that evidence into a concrete improvement. Use when asked to review, evaluate, diagnose, or "analyse the conversation" for an agent — the console's Analyze button lands here with a transcript pre-filled. Grounds any persona change in what users said and how the agent replied, then feeds it back into the get → update → publish loop.
- ▌ Frontend Development · clode-labsBuild modern frontend applications using React, Vue, or vanilla JavaScript. Use this skill for creating UI components, pages, forms, and interactive web interfaces with proper styling, accessibility, and responsive design.
- ▌ Intervix MCP · clode-labsIntervix recruiter tools via intervix's MCP server. Create and manage job postings, send interview invites (single or bulk), read finished interview status / transcripts / recordings / proctor events, and manage interviewer agents — all scoped to the caller's organization.
- ▌ Aramb Browser · clode-labsThe way to touch JS-rendered, authenticated, or visually-inspected web content — every URL visit, scrape, search engine query, form fill, screenshot, JS evaluation, or live data fetch on a rendered/restricted site goes through this skill. For those sites do NOT use WebSearch, WebFetch, curl, wget, or HTTP libraries — they use the datacenter UA, can't execute JS, and return SPA HTML or 403 from any half-restricted site. Public/static content (GitHub repos & raw files, plain pages, JSON/APIs) is the opposite case — fetch it with curl/git clone/WebFetch, not the browser (see the Fetch hierarchy). Fork of chrome-devtools-mcp; standard CDP actions all work.
- ▌ Skill Creator · clode-labsGuide for creating effective aramb skills. Use this skill when users want to create a new skill or update an existing skill that extends agent capabilities with specialized knowledge, workflows, or domain expertise for the aramb orchestration system.
- ▌ Aramb Metadata · clode-labsGenerate or update aramb.toml configuration file by analyzing docker-compose files or codebase. Use this skill when you need to create project metadata, service configurations, or environment setup for aramb-orchestrated projects.
- ▌ Aramb Toolkits · clode-labsThe SINGLE surface for third-party toolkits (Composio integrations) — discover, inspect, and EXECUTE tools, manage connections, read the event- trigger catalog, and mint github credentials. All via `aramb_mcp.toolkits_*`. Discover a tool (search → get_schema) then run it (execute). Check/list/connect accounts. For GitHub: execute GITHUB_GET_GIT_CREDENTIAL for a token, then use native git/gh. This skill fully replaces the retired `composio-cli` — there is no `composio` CLI; everything is an `aramb_mcp.toolkits_*` MCP call.
- ▌ Checker Prompt · clode-labsCraft a precise checker_prompt for external validation — specifies what the checker agent must verify
- ▌ Aramb Knowledge · clode-labsRetrieve facts from your agent's knowledge base via semantic search (the aramb_mcp server, knowledge_search tool). Use it to answer any factual question — policies, prices, limits, dates, procedures, product or domain details — from the documents uploaded to THIS agent, instead of guessing or answering from memory.
- ▌ Aramb Workflows · clode-labsMCP toolkit for workflows (aramb_mcp.workflows_*). Use these to create, read, update, schedule, and consolidate workflows, plus to update workflow run steps when dispatched as a workflow step agent.
- ▌ Backend Planner · clode-labsPlan complex backend work by analyzing requirements and creating executable task sequences. Use this skill when breaking down API features, service architectures, database changes, or backend integrations into development and QA tasks with self-validation.
- ▌ Backend Testing · clode-labsAPI testing protocol — start the stack, discover endpoints, test each one with assertions, and report results using the verdict protocol. Use when: (1) testing API endpoints after development, (2) verifying service health and endpoint correctness, (3) running regression tests on backend services. NOT for: frontend/UI testing, deployment, or code changes.
- ▌ Create Workflow · clode-labsAuthor a brand-new workflow definition (no workflow exists yet). Two dispatch channels: (1) task dispatch — the master consolidates the application's user tasks into a reusable workflow; (2) chat dispatch — author from the user's explicit description or the work done so far in this conversation (the master in team mode, or the solo agent in solo mode). Each node gets an agent for its role the SAME way in solo and team — reuse an existing agent, or author the sub-agent's full spec INLINE on the workflow via `agent_specs`; mode affects only close-out (chat vs tasks), never how nodes get their agents. Use when: asked to create a workflow from prior tasks, "build a workflow that…", "create a workflow based on the work done so far in this chat", or told to use the create-workflow skill. NOT for: editing an existing workflow (use update-workflow), polishing a template import (use import-workflow), executing or scheduling workflows.
- ▌ Import Workflow · clode-labsEnrich a pre-created workflow draft for the current application — fetch the draft the platform already built from the template, polish its node prompts using the user's wizard answers, save the polish, and (if asked) set its schedule. The platform then auto-publishes and runs it. Use when the dispatch contains a <template-import> block in the extra-system-prompt. NOT for: consolidating completed tasks (use create-workflow), editing an arbitrary existing workflow (use update-workflow), or executing a workflow.
- ▌ Update Workflow · clode-labsRegenerate / refresh / tweak an existing workflow's definition. Works in two dispatch modes: (1) task dispatch — master regenerates from the application's user tasks (any status) plus any change request (system task with purpose=update-workflow); (2) chat dispatch — solo applies an explicit change request ("add a Slack DM step", "remove the synth node") or consolidates the work done so far in this chat. Use when: dispatched as a system task with purpose=update-workflow, OR when the user asks to refresh / update / regenerate / change the workflow. NOT for: creating a workflow from scratch (use create-workflow), polishing a template import (use import-workflow), executing a workflow, or editing schedules (use schedule-workflow).
- ▌ Agent Archetypes · clode-labsAn OPTIONAL shortcut library of a few common agent shapes for the Architect. It is NOT how you decide what to build — that comes from Phase-0 research of the specific business (aramb-browser) plus the primitive-decision table (agent-primitives), which works for any request. Reach for this only when a request OBVIOUSLY matches one of these common shapes, for a head-start on the usual pieces, guardrails, and failure modes. Most real requests won't fit cleanly — that's expected; never force one.
- ▌ Aramb Deployment · clode-labsBuild and deploy applications using aramb-cli. Covers aramb-cli installation, deployment mode detection, env var/secret filling, service creation, local Docker builds, and deployment orchestration. Use for all deployment tasks. TOML generation is handled by the aramb-toml skill.
- ▌ Backend Critique · clode-labsQA skill for backend work. Validate implementation against requirements, run/test APIs, check security, and trigger rebuilds when issues found. Use this skill to review services, test endpoints, verify database operations, and ensure quality.
- ▌ Frontend Planner · clode-labsPlan complex frontend work by analyzing requirements and creating executable task sequences. Use this skill when breaking down UI features, component architectures, or frontend integrations into development and QA tasks with self-validation.
- ▌ Frontend Testing · clode-labsPlaywright-based UI testing protocol — start the stack, navigate pages, fill forms, click buttons, verify renders, capture screenshots on failure, handle waits and loading states. Use when: (1) testing frontend UI flows after development, (2) verifying pages render and interactions work, (3) regression testing UI components. NOT for: API testing, deployment, or code changes.
- ▌ Aramb UI · clode-labsMCP toolkit for driving the console UI (aramb_mcp.ui_*). Use these to move the user's view to a specific console page (navigate) and to read a page's structured layout — its buttons, tabs, and where they sit — so you can point the user at the exact on-screen control. You live in the console side panel, so put the user in front of the right screen instead of only describing where to click.
- ▌ Chil MCP · clode-labsChat tools via chil's MCP server. Post, reply, DM, ask, read channels and threads, list a user's channels — on whatever chat platform the caller's org has installed (Slack today; MS Teams / Discord later).
- ▌ Planning · clode-labsInteractive planning for complex requests — QnA, plan file, structured approval.
- ▌ Vault MCP · clode-labsYour own secure secret vault via the aramb_mcp server (vault_* tools). Store, fetch, list, and delete your project-scoped secrets (API tokens, keys, credentials) — kept in a real secrets manager, never in chat, files, or git. Also create write-only placeholder secrets for the USER to fill with a credential you must not see. Use when you need to save or retrieve a credential for yourself, or to prompt the user for one. NOT for GitHub repository secrets (that is `gh secret set`).
- ▌ Aramb Chat · clode-labsMCP toolkit for the chat surface (aramb_mcp.chat_*). Use these to send messages, ask the user questions, alert them, surface artifacts (files & URLs) outside a task, work with linked GitHub repos, drive the planning flow, and read the current chat's mode (solo vs team).
- ▌ Aramb Toml · clode-labsGenerate and update aramb.toml configuration files for deploying applications on aramb. Covers codebase analysis, service type mapping, variable/secret classification, TOML structure, and validation. Use this skill whenever aramb.toml needs to be created or updated.
- ▌ Indeed MCP · clode-labsIndeed job search via the indeed-mcp server. Search live Indeed job postings with rich filters (location, country, remote, job type, recency) and look up an employer's public profile plus its open roles. Anonymous public data — no candidate login, no resumes.
- ▌ Agent Tests · clode-labsAuthor and run scripted, repeatable test suites against an agent persona (aramb_mcp.agents_test_*). Use to prove a persona behaves — a test is a scripted multi-turn conversation plus a success_condition, and a run replays the user turns against the live (or draft) config and hands back the transcript to judge. This is the ONLY supported way to test a persona; never hand-roll a side conversation or write fixtures to disk. Every test MUST script at least 3 user turns.
- ▌ Aramb Email · clode-labsYour own email address (inbox + outbox) via the aramb_mcp server (email_* tools). SEND mail as yourself, READ the mail delivered to you, and REPLY to it, threaded. Your from-address is your own agent address and is stamped server-side — you cannot send as anyone else. Use for sign-ups / confirmations / magic-links that need a real inbox, for corresponding with a person by email, and for handling mail forwarded to you. All via `aramb_mcp.email_*`.
- ▌ Aramb Phone · clode-labsMCP toolkit for placing a real outbound phone call (aramb_mcp.phone_call). Use it to call a phone number from a published agent's claimed number; the agent speaks with whoever answers. Use when the user asks you to phone someone, confirm something by voice, or leave a spoken message. NOT for chat messages or notifications — that is aramb_chat.
- ▌ Aramb Tasks · clode-labsMCP toolkit for the task lifecycle (aramb_mcp.tasks_*). Use these to create, update, and list tasks. The task lifecycle is the team-mode primitive; this toolkit is hidden in solo mode by tools/list filtering.
- ▌ Aramb Voice · clode-labsMCP toolkit for voice notes (aramb_mcp.voice_transcribe / aramb_mcp.voice_synthesize). Use voice_transcribe to read an incoming voice note (audio file → text), and voice_synthesize to speak a reply (text → audio) that you then hand to aramb_chat.deliver_artifacts as a downloadable/playable attachment. Use for chat surfaces that exchange voice messages (Slack, WhatsApp, Telegram, and the like). NOT for placing phone calls — that is aramb_phone.
- ▌ E2e Testing · clode-labsEnd-to-end testing protocol — test complete user journeys across services, verify data flows through API, UI, and database as a coherent system. Use when: (1) verifying full user journeys after backend and frontend pass individual tests, (2) testing cross-service data flows, (3) regression testing complete workflows. NOT for: individual endpoint testing (use backend-testing), individual UI testing (use frontend-testing), or code changes.
- ▌ Aramb Agents · clode-labsMCP toolkit for the organization's aramb agents (aramb_mcp.agents_*). Use to create an agent once a persona design is settled, and to inspect, revise (draft), and publish existing agents. To read an agent's real conversations when evaluating it use the analyse-conversation skill; to author and run scripted test suites use the agent-tests skill (both are aramb_mcp.agents_*). NOT for provisioning workflow-node sub-agents — that is the create-agent skill.
- ▌ Aramb Expose · clode-labsExpose local services publicly via aramb expose tunnels (proxy.clode.space URLs). Use when: making a local service accessible via a public URL for preview/demo purposes. Creates named tunnel clients that persist in the registry and can be reused across deployments. NOT for: production deployments or exposing databases/caches.
- ▌ Aramb Skills · clode-labsSearch, inspect, and download skills from the Skills Registry using the aramb-skills CLI. Available natively at /usr/local/bin/aramb-skills inside all agent runtime containers — no install needed. Use when: discovering available skills before creating agents, checking if a skill exists in the registry, downloading a skill into an agent workspace, or listing skills by category or tag.
- ▌ Create Agent · clode-labs bundleCreate new agents with tailored workspace files (IDENTITY.md, SOUL.md, AGENTS.md), custom skills, and CLI registration. Use when: (1) setting up a new isolated agent for a specific role (coder, tester, deployer, researcher, etc.), (2) scaffolding an agent workspace with role-specific persona and instructions, (3) registering an agent via `benji agent create`. NOT for: editing existing agent files (just edit them directly), or creating skills without an agent (write SKILL.md files directly).
- ▌ Dev Workflow · clode-labsDevelopment workflow protocol — branching, committing, project structure, and docker-compose conventions. Use when: (1) starting a new project or feature, (2) making commits, (3) setting up project structure with docker-compose, (4) completing a development task. NOT for: deployment, CI/CD pipeline configuration, or infrastructure changes.
- ▌ Juno · clode-labsContext memory — store and retrieve patterns, gotchas, and insights across sessions.
- ▌ Solo · clode-labsTools and conventions for the solo (direct-execution) agent. Communication, git, preview URLs, workflow authoring, and scheduling. No platform task surface (aramb_mcp.tasks_* is filtered out in solo mode); sub-agents are allowed for workflow nodes.