跳到正文
cccskills
免费在 GitHub 公开

posthog-automation

Automate PostHog tasks via Rube MCP (Composio): events, feature flags, projects, user profiles, annotations. Always search tools first for current schemas.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(1)

  • SKILL.md8.0 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

PostHog Automation via Rube MCP

Automate PostHog product analytics and feature flag management through Composio's PostHog toolkit via Rube MCP.

Toolkit docs: composio.dev/toolkits/posthog

Prerequisites

  • Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
  • Active PostHog connection via RUBE_MANAGE_CONNECTIONS with toolkit posthog
  • Always call RUBE_SEARCH_TOOLS first to get current tool schemas

Setup

Get Rube MCP: Add https://rube.app/mcp as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.

  1. Verify Rube MCP is available by confirming RUBE_SEARCH_TOOLS responds
  2. Call RUBE_MANAGE_CONNECTIONS with toolkit posthog
  3. If connection is not ACTIVE, follow the returned auth link to complete PostHog authentication
  4. Confirm connection status shows ACTIVE before running any workflows

Core Workflows

1. Capture Events

When to use: User wants to send event data to PostHog for analytics tracking

Tool sequence:

  1. POSTHOG_CAPTURE_EVENT - Send one or more events to PostHog [Required]

Key parameters:

  • event: Event name (e.g., '$pageview', 'user_signed_up', 'purchase_completed')
  • distinct_id: Unique user identifier (required)
  • properties: Object with event-specific properties
  • timestamp: ISO 8601 timestamp (optional; defaults to server time)

Pitfalls:

  • distinct_id is required for every event; identifies the user/device
  • PostHog system events use $ prefix (e.g., '$pageview', '$identify')
  • Custom events should NOT use the $ prefix
  • Properties are freeform; maintain consistent schemas across events
  • Events are processed asynchronously; ingestion delay is typically seconds

2. List and Filter Events

When to use: User wants to browse or search through captured events

Tool sequence:

  1. POSTHOG_LIST_AND_FILTER_PROJECT_EVENTS - Query events with filters [Required]

Key parameters:

  • project_id: PostHog project ID (required)
  • event: Filter by event name
  • person_id: Filter by person ID
  • after: Events after this ISO 8601 timestamp
  • before: Events before this ISO 8601 timestamp
  • limit: Maximum events to return
  • offset: Pagination offset

Pitfalls:

  • project_id is required; resolve via LIST_PROJECTS first
  • Date filters use ISO 8601 format (e.g., '2024-01-15T00:00:00Z')
  • Large event volumes require pagination; use offset and limit
  • Results are returned in reverse chronological order by default
  • Event properties are nested; parse carefully

3. Manage Feature Flags

When to use: User wants to create, view, or manage feature flags

Tool sequence:

  1. POSTHOG_LIST_AND_MANAGE_PROJECT_FEATURE_FLAGS - List existing feature flags [Required]
  2. POSTHOG_RETRIEVE_FEATURE_FLAG_DETAILS - Get detailed flag configuration [Optional]
  3. POSTHOG_CREATE_FEATURE_FLAGS_FOR_PROJECT - Create a new feature flag [Optional]

Key parameters:

  • For listing: project_id (required)
  • For details: project_id, id (feature flag ID)
  • For creation:
    • project_id: Target project
    • key: Flag key (e.g., 'new-dashboard-beta')
    • name: Human-readable name
    • filters: Targeting rules and rollout percentage
    • active: Whether the flag is enabled

Pitfalls:

  • Feature flag key must be unique within a project
  • Flag keys should use kebab-case (e.g., 'my-feature-flag')
  • filters define targeting groups with properties and rollout percentages
  • Creating a flag with active: true immediately enables it for matching users
  • Flag changes take effect within seconds due to PostHog's polling mechanism

4. Manage Projects

When to use: User wants to list or inspect PostHog projects and organizations

Tool sequence:

  1. POSTHOG_LIST_PROJECTS_IN_ORGANIZATION_WITH_PAGINATION - List all projects [Required]

Key parameters:

  • organization_id: Organization identifier (may be optional depending on auth)
  • limit: Number of results per page
  • offset: Pagination offset

Pitfalls:

  • Project IDs are numeric; used as parameters in most other endpoints
  • Organization ID may be required; check your PostHog setup
  • Pagination is offset-based; iterate until results are empty
  • Project settings include API keys and configuration details

5. User Profile and Authentication

When to use: User wants to check current user details or verify API access

Tool sequence:

  1. POSTHOG_WHOAMI - Get current API user information [Optional]
  2. POSTHOG_RETRIEVE_CURRENT_USER_PROFILE - Get detailed user profile [Optional]

Key parameters:

  • No required parameters for either call
  • Returns current authenticated user's details, permissions, and organization info

Pitfalls:

  • WHOAMI is a lightweight check; use for verifying API connectivity
  • User profile includes organization membership and permissions
  • These endpoints confirm the API key's access level and scope

Common Patterns

ID Resolution

Organization -> Project ID:

1. Call POSTHOG_LIST_PROJECTS_IN_ORGANIZATION_WITH_PAGINATION
2. Find project by name in results
3. Extract id (numeric) for use in other endpoints

Feature flag name -> Flag ID:

1. Call POSTHOG_LIST_AND_MANAGE_PROJECT_FEATURE_FLAGS with project_id
2. Find flag by key or name
3. Extract id for detailed operations

Feature Flag Targeting

Feature flags support sophisticated targeting:

{
  "filters": {
    "groups": [
      {
        "properties": [
          {"key": "email", "value": "@company.com", "operator": "icontains"}
        ],
        "rollout_percentage": 100
      },
      {
        "properties": [],
        "rollout_percentage": 10
      }
    ]
  }
}
  • Groups are evaluated in order; first matching group determines the rollout
  • Properties filter users by their traits
  • Rollout percentage determines what fraction of matching users see the flag

Pagination

  • Events: Use offset and limit (offset-based)
  • Feature flags: Use offset and limit (offset-based)
  • Projects: Use offset and limit (offset-based)
  • Continue until results array is empty or smaller than limit

Known Pitfalls

Project IDs:

  • Required for most API endpoints
  • Always resolve project names to numeric IDs first
  • Multiple projects can exist in one organization

Event Naming:

  • System events use $ prefix ($pageview, $identify, $autocapture)
  • Custom events should NOT use $ prefix
  • Event names are case-sensitive; maintain consistency

Feature Flags:

  • Flag keys must be unique within a project
  • Use kebab-case for flag keys
  • Changes propagate within seconds
  • Deleting a flag is permanent; consider disabling instead

Rate Limits:

  • Event ingestion has throughput limits
  • Batch events where possible for efficiency
  • API endpoints have per-minute rate limits

Response Parsing:

  • Response data may be nested under data or results key
  • Paginated responses include count, next, previous fields
  • Event properties are nested objects; access carefully
  • Parse defensively with fallbacks for optional fields

Quick Reference

TaskTool SlugKey Params
Capture eventPOSTHOG_CAPTURE_EVENTevent, distinct_id, properties
List eventsPOSTHOG_LIST_AND_FILTER_PROJECT_EVENTSproject_id, event, after, before
List feature flagsPOSTHOG_LIST_AND_MANAGE_PROJECT_FEATURE_FLAGSproject_id
Get flag detailsPOSTHOG_RETRIEVE_FEATURE_FLAG_DETAILSproject_id, id
Create flagPOSTHOG_CREATE_FEATURE_FLAGS_FOR_PROJECTproject_id, key, filters
List projectsPOSTHOG_LIST_PROJECTS_IN_ORGANIZATION_WITH_PAGINATIONorganization_id
Who am IPOSTHOG_WHOAMI(none)
User profilePOSTHOG_RETRIEVE_CURRENT_USER_PROFILE(none)

Powered by Composio

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Handle ActiveCampaign contact and marketing tasks through Rube MCP, a connection for agent tools. Find or create contacts, manage tags and subscriptions, enroll contacts in existing automated workflows, and create follow-up tasks.

中文简介正在准备中。当前显示英语简介。

  • Finding or registering contacts
  • Adding or removing contact tags
  • Subscribing contacts to lists or unsubscribing them
davepoon/buildwithclaude36662026年10月20日 更新

Research competitors’ live Meta and Google ads to identify opening hooks, offers, calls to action, and landing pages. Receive a sourced creative brief and up to three test ideas adapted to your product.

中文简介正在准备中。当前显示英语简介。

  • Find angles for new ad copy
  • Plan creative tests using competitor ads
  • Choose a landing page message
davepoon/buildwithclaude36662026年10月20日 更新

Analytics your AI agent can actually use. Track, analyze, run A/B experiments, and optimize across all your projects via CLI. Includes a growth playbook so your agent knows HOW to grow, not just what to track.

中文简介正在准备中。当前显示原文说明。

davepoon/buildwithclaude36662026年10月20日 更新

Teaches when to recall from long-term memory before acting and when to save durable decisions, corrections and failures afterwards. Use when a memory tool or MCP memory server is connected but the agent is not using it consistently, when the user complains that the assistant forgets preferences, conventions or past decisions between sessions, or when setting up persistent memory for a project. Works with any memory backend: a folder of Markdown files, a local MCP server, or a managed service.

中文简介正在准备中。当前显示原文说明。

davepoon/buildwithclaude36662026年10月20日 更新

Audit local AI coding-agent sessions with agenttrace for cost, tokens, tool failures, latency, anomalies, health, diffs, and CI gates.

中文简介正在准备中。当前显示原文说明。

davepoon/buildwithclaude36662026年10月20日 更新

Checks which private files (.env, keys, secrets) Claude Code and its subagents reached in this project, whether a value came into an agent's context, and which key to change, by running the open source agentwhy CLI over the session records Claude Code already keeps on disk. Use when the user asks whether the agent or a subagent read their .env or other secrets, which private files a session touched, whether a key leaked into the conversation, or which key to rotate. Runs locally and never prints a secret value.

中文简介正在准备中。当前显示原文说明。

davepoon/buildwithclaude36662026年10月20日 更新

查看 davepoon 的全部技能

报告此技能的问题