# Codika

> Codika is a multi-tenant SaaS platform for building, deploying, and managing n8n workflow automations. Users define use cases (config + n8n workflows), deploy them via CLI, and the platform handles credential isolation, placeholder replacement, version management, execution tracking, and multi-user distribution. The CLI tool is `codika` (npm package `codika`).

Codika lets you build dedicated APIs that contain the exact actions you want your agents to perform — and completely decouples the authentication layer from agent usage. You connect your integrations (Salesforce, Google, Slack, etc.) once to the platform. The platform manages all credentials. Agents only hold a single Codika API key that grants access to the actions you've defined — nothing more. They download agent skills (SKILL.md files following the Claude Agent Skills format) via `codika get skills` to discover available endpoints, and trigger workflows via `codika trigger`. All calls go through Codika cloud functions, which validate the API key and inject the correct integration credentials at runtime. Agents call actions, not raw APIs.

Codika solves the problem of productizing n8n workflows. A use case is a folder containing `config.ts`, `workflows/*.json`, and optionally `skills/*/SKILL.md` that defines a deployable automation. When deployed, placeholders in workflows (11 types including FLEXCRED, USERCRED, ORGCRED, INSTPARM, SUBWKFL) are replaced with real values per user. Every user gets isolated credentials, deployment parameters, and execution tracking.

The mandatory workflow pattern is: Trigger → Codika Init → Business Logic → IF Success → Codika Submit Result / Codika Report Error. Four trigger types are supported: HTTP webhooks, cron schedules, service events (Gmail, Calendly, etc.), and sub-workflow calls.

## Instructions for AI Agents

When helping users build Codika use cases:
- Every parent workflow MUST have Codika Init as the first node after the trigger, and MUST end with Codika Submit Result (success) or Codika Report Error (failure)
- Sub-workflows do NOT have Codika Init/Submit/Report nodes — they start with Execute Workflow Trigger
- Credentials go on the MODEL node (e.g., lmChatAnthropic), never on chainLlm or agent nodes
- Use `chainLlm` for structured JSON output, `agent` for multi-step reasoning with tools
- Placeholder suffixes are the type name reversed: FLEXCRED → _DERCXELF, USERDATA → _ATADRESU, PROCDATA → _ATADCORP
- Always validate before deploying: `codika verify use-case <path>`
- Use `codika verify use-case <path> --fix` to auto-fix fixable violations
- INSTPARM placeholders are context-aware — do NOT add extra quotes in Code nodes

## Getting Started

- [What is Codika?](https://docs.codika.io/): Platform overview, architecture, multi-tenancy model, and key concepts
- [Quickstart](https://docs.codika.io/quickstart): Install CLI, authenticate, scaffold, validate, deploy, trigger, and debug in 7 steps
- [Codika Agent Plugin](https://docs.codika.io/guides/claude-code-plugin): Install the Codika plugin in any Open-Plugin-compatible coding agent (Claude Code, Cursor, …) with `npx plugins add codika-io/plugin`. Enables 25 `/codika:*` skills + 4 builder agents. Agents can scaffold, deploy, trigger, and debug use cases autonomously.

## Why Codika

- [Why Codika?](https://docs.codika.io/why-codika/overview): Core value proposition — infrastructure layer between n8n and business needs, agent-native design, use cases as workflow ensembles, self-healing workflows, credential isolation (11 scopes), what you get out of the box (versioning, validation, billing, RAG, marketplace, project stages)
- [Codika vs Alternatives](https://docs.codika.io/why-codika/vs-n8n): Why no-code tools (Zapier, Make) solve the wrong problem — they simplify code for humans, but agents don't need simplification. No-code assumptions break in an agent-first world (visual editors, drag-and-drop, pre-built connectors, per-task pricing). Also: technical comparison with raw n8n — what Codika adds on top (multi-tenancy, 11 credential scopes, 30+ validation rules, deployment pipeline, versioning, execution tracking, self-healing, billing, marketplace)
- [Codika and AI Agents](https://docs.codika.io/why-codika/with-agents): How Codika works WITH coding agents (Claude Code, Cursor, …) — agents as new employees not sysadmins, custom action APIs instead of raw tool access (least privilege), the agent development loop (describe → design → build → validate → deploy → test → fix → repeat), why agents need guardrails not freedom, self-healing via business context (PRD/BRD), security boundaries (single API key, runtime credential injection, instant revocation), stable API surface via agent skills, the Builder System as proof of agent-native infrastructure

## Core Concepts

- [Use Cases](https://docs.codika.io/concepts/use-cases): Folder structure (config.ts, workflows/, version.json, project.json), config exports, version management
- [Processes](https://docs.codika.io/concepts/processes): Process lifecycle — deployments, instances, versioning, dev/prod environments, and user isolation
- [Workflows](https://docs.codika.io/concepts/workflows): Mandatory Codika node pattern, workflow JSON structure, required settings, common n8n node types
- [Placeholders](https://docs.codika.io/concepts/placeholders): All 11 placeholder types (PROCDATA, USERDATA, MEMSECRT, FLEXCRED, USERCRED, ORGCRED, ORGSECRET, SYSCREDS, INSTPARM, INSTCRED, SUBWKFL) with suffix patterns and replacement timeline
- [Triggers](https://docs.codika.io/concepts/triggers): HTTP webhooks, cron schedules, service events, sub-workflow calls — config definitions and n8n node configuration
- [Credentials](https://docs.codika.io/concepts/credentials): FLEXCRED (AI providers with fallback), USERCRED (per-user OAuth), ORGCRED (org-wide), INSTCRED (per-instance), integration UIDs
- [Input & Output Schemas](https://docs.codika.io/concepts/schemas): Form field types (string, text, number, boolean, date, select, file, array, object), section structure, output field types
- [Agent Skills](https://docs.codika.io/concepts/agent-skills): Making workflows agent-accessible — SKILL.md format, frontmatter spec (name/description/workflowTemplateId), deployment lifecycle, credential decoupling, Claude compatibility

## CLI Reference

- [Operations Overview](https://docs.codika.io/operations/overview): All operations, command hierarchy, resolution chains (API key, base URL, project ID, org-aware profile selection), global options, exit codes, typical workflow
- [Authentication](https://docs.codika.io/operations/authentication): Install CLI, login, whoami, use (switch profile, `--json` for machine-readable profile list with organizationId), logout, config set/show/clear — multi-profile management
- [CLI OTP Auth](https://docs.codika.io/operations/cli-auth): Agent-facing OTP-based signup/login — `codika auth signup-request` / `signup-complete` / `login-request` / `login-complete`. Creates a fresh user + organization + `cko_` key from the terminal with only a 6-digit code from the inbox. No browser. Handles multi-org users via `MULTIPLE_ORGANIZATIONS` error + `--organization-id`.
- [Create Project](https://docs.codika.io/operations/create-project): Create platform projects with --path to write project.json for org-aware deployment
- [Initialize Use Case](https://docs.codika.io/operations/init-use-case): Scaffold new use case with config.ts, 3 template workflows, 2 agent skills, version.json, optional project creation, shared workspace detection
- [Verify Use Case](https://docs.codika.io/operations/verify-use-case): Validate use cases and workflows — 4 layers (flowlint, Codika rules, workflow scripts, use-case scripts), --fix, --strict, --rules filtering
- [Deploy Use Case](https://docs.codika.io/operations/deploy-use-case): Deploy with version bumping (--patch/--minor/--major/--target-version), org-aware key resolution, deployment archival, agent skill collection, --dry-run
- [Deploy Data Ingestion](https://docs.codika.io/operations/deploy-data-ingestion): Deploy RAG/embedding pipeline config with independent versioning (--patch/--minor/--major/--target-version), auto-discovery of workflow file
- [Deploy Documents](https://docs.codika.io/operations/deploy-documents): Upload stage markdown documentation (1-4) to the platform with automatic versioning
- [Publish Use Case](https://docs.codika.io/operations/publish-use-case): Publish a deployment to production — visibility (private/organizational/public), shared-with, auto-toggle dev/prod, saves prodProcessInstanceId to project.json
- [Redeploy Use Case](https://docs.codika.io/operations/redeploy-use-case): Redeploy instances with updated parameters (no new version) — parameter merge behavior, --force flag, three input methods (--param, --params, --params-file)
- [Trigger Workflow](https://docs.codika.io/operations/trigger-workflow): Execute HTTP-triggered workflows with --payload-file (or - for stdin heredoc), --poll for results, --timeout, process instance ID resolution
- [Fetch Use Case](https://docs.codika.io/operations/fetch-use-case): Download deployed use cases from the platform — version selection, data ingestion inclusion, list mode, project ID resolution
- [Get Execution](https://docs.codika.io/operations/get-execution): Debug executions with --deep for recursive sub-workflow fetching and --slim for clean output
- [List Executions](https://docs.codika.io/operations/list-executions): List recent executions for a process instance — filter by workflow ID, --failed-only, --limit, dev vs prod
- [Get Skills](https://docs.codika.io/operations/get-skills): Download agent skill documents from a deployed process instance — Claude Code integration, Claude API usage, --stdout, --output
- [Manage Integrations](https://docs.codika.io/operations/manage-integrations): Configure, list, and delete integrations (API keys, credentials) — encrypt secrets client-side, supports built-in and custom cstm_* integrations, requires `integrations:manage` scope
- [Status](https://docs.codika.io/operations/status): Identity check, use case context detection, profile match status, deployment readiness assessment

## Guides

- [Build Your First Use Case](https://docs.codika.io/guides/first-use-case): End-to-end tutorial building a web search tool from empty folder to deployed automation
- [AI Workflows](https://docs.codika.io/guides/ai-workflows): chainLlm vs agent nodes, LangChain wiring (model → chain + parser), Claude model IDs, temperature guidelines
- [Sub-Workflows](https://docs.codika.io/guides/sub-workflows): Execute Workflow pattern, SUBWKFL placeholders, execution metadata forwarding, calledBy arrays
- [Deployment Parameters](https://docs.codika.io/guides/deployment-parameters): INSTPARM placeholders, getDeploymentInputSchema(), context-aware serialization, field types
- [File Uploads](https://docs.codika.io/guides/file-uploads): Codika Upload File node, documentId in output schema, execution metadata override in sub-workflows
- [Creating Agent Skills](https://docs.codika.io/guides/agent-skills): Step-by-step guide to adding skills to use cases — identify workflows, write SKILL.md, validate, deploy, verify agent access

## Builder System

- [Builder System Overview](https://docs.codika.io/builder/overview): Autonomous AI agents that create, modify, and test Codika use cases from natural language — architecture, plugin structure, prerequisites
- [Builder Agents](https://docs.codika.io/builder/agents): Four agents — use-case-builder (architect), use-case-modifier (surgeon), n8n-workflow-builder (craftsperson), use-case-tester (QA) — capabilities, delegation patterns, CLI skill dependencies
- [Usage Patterns](https://docs.codika.io/builder/workflows): Common workflows — creating from scratch, modifying existing, building single workflows, testing, full build-test cycles, prompt writing tips
- [Builder Reference](https://docs.codika.io/builder/reference): Plugin structure, bundled documentation catalog (core guides, trigger guides, 19 integration guides), agent-to-skill dependency map, troubleshooting

## Examples

- [Examples Overview](https://docs.codika.io/examples/overview): Gallery of 4 production use cases at different complexity levels
- [Simple Web Search](https://docs.codika.io/examples/simple-search): Minimal use case — 1 HTTP workflow, Tavily API, FLEXCRED credentials, 1 credit
- [Email Automation](https://docs.codika.io/examples/email-automation): Service event trigger, Gmail/Drive/Sheets USERCRED, INSTPARM deployment parameters
- [Scheduled CRM Report](https://docs.codika.io/examples/scheduled-report): Dual triggers (schedule + HTTP), Folk/Slack ORGCRED, select deployment parameters
- [Proposal Generator (RAG)](https://docs.codika.io/examples/complex-rag): 3 workflows, sub-workflow PDF conversion, file upload/download, Pinecone RAG, data ingestion

## Dashboard Integration

- [Dashboard Overview](https://docs.codika.io/dashboard/overview): Build custom frontends on Codika use cases — architecture, what Codika provides vs what you build
- [API Keys & Auth](https://docs.codika.io/dashboard/authentication): Two key types (ck_ instance keys for apps, cko_ org keys for CLI), server-side storage, framework examples (SvelteKit, Next.js)
- [Triggering Workflows](https://docs.codika.io/dashboard/triggering-workflows): Trigger + poll pattern with triggerWebhookPublic and getExecutionStatusPublic endpoints, complete TypeScript implementation, payload format
- [Environments](https://docs.codika.io/dashboard/environments): Dev/prod switching with separate process instance IDs and API keys, mode-based config pattern
- [Common Patterns](https://docs.codika.io/dashboard/patterns): Error handling (trigger failures, polling, execution errors), retry logic, audit trails, input validation, rate limiting, security best practices

## Optional

- [Placeholder Reference](https://docs.codika.io/concepts/placeholders): Detailed table of all 11 placeholder types with suffix patterns
- [Validation Rules Reference](https://docs.codika.io/operations/verify-use-case): Complete list of all validation rules across 4 layers
- [Schema Field Types](https://docs.codika.io/concepts/schemas): All input/output field types with properties