Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 1 addition & 3 deletions contents/docs/ai-observability/skills.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -13,9 +13,7 @@ PostHog ships agent skills for AI Observability that teach your coding agent how

1. Make sure there's LLM data to investigate – if you haven't instrumented yet, [install a PostHog SDK or provider integration](/docs/ai-observability/installation).

2. [Connect the PostHog MCP server](/docs/ai-observability/surfaces/mcp). (The AI plugin in step 3 does this for you.)

3. Install the [PostHog AI plugin](https://github.com/PostHog/ai-plugin) – it bundles every PostHog skill and keeps them up to date. Its README covers installation in Claude Code, Codex, Cursor, and Gemini CLI.
2. [Connect the PostHog MCP server](/docs/ai-observability/surfaces/mcp). The skills ship with the MCP server, so there's nothing extra to install: your agent loads the one it needs when a task calls for it.

<CalloutBox icon="IconInfo" title="Want the skills directly?" type="fyi">

Expand Down
2 changes: 1 addition & 1 deletion contents/docs/data-pipelines/migrate-off-fivetran.md
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ Prefer to watch how it works? Follow the instruction video below:
You need:

- An MCP client (like Claude Code, Cursor, or Claude Desktop) with the [PostHog MCP server](/docs/model-context-protocol) installed
- The `migrating-from-fivetran` agent skill. It ships with the [PostHog MCP server](/docs/model-context-protocol), so there's nothing extra to install: ask your agent to move your Fivetran setup and it loads the skill. You can also get it from the [PostHog AI plugin](https://github.com/PostHog/ai-plugin), or download `skills.zip` from the [latest skills release](https://github.com/PostHog/posthog/releases/tag/agent-skills-latest). See [Skills](/docs/skills) for more.
- The `migrating-from-fivetran` agent skill. It ships with the [PostHog MCP server](/docs/model-context-protocol), so there's nothing extra to install: ask your agent to move your Fivetran setup and it loads the skill. You can also download `skills.zip` from the [latest skills release](https://github.com/PostHog/posthog/releases/tag/agent-skills-latest). See [Skills](/docs/skills) for more.
- A way to read your Fivetran account, using either:
- A Fivetran API key and secret (recommended). The key can be read-only. You can generate these in Fivetran under your user settings ([Fivetran docs](https://fivetran.com/docs/rest-api/getting-started))
- The Fivetran MCP server. All of your Fivetran setup goes through the model, so this uses more tokens, especially if your setup is large
Expand Down
2 changes: 1 addition & 1 deletion contents/docs/replay-vision/mcp.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -121,7 +121,7 @@ Each tool needs the same access as the matching action in the app: read tools ne

## Agent skills

PostHog also publishes three Replay Vision [agent skills](https://github.com/PostHog/skills): instructions that teach an agent how to use these tools well. They ship in the [PostHog AI plugin](https://github.com/PostHog/ai-plugin) for Claude Code, Codex, Cursor, and Gemini CLI, so an agent with the plugin installed picks them up when the task calls for them.
PostHog also publishes three Replay Vision [agent skills](https://github.com/PostHog/skills): instructions that teach an agent how to use these tools well. They ship with the [PostHog MCP server](/docs/model-context-protocol), so a connected agent loads them when the task calls for them.

- `designing-replay-vision-scanners` – designs a scanner that produces trustworthy observations: one visible question, the right type, a query that selects only sessions able to answer it, and a prompt that asks for on-screen proof.
- `creating-replay-vision-scanners` – creates a scanner and sizes it safely, estimating its spend and checking the budget before it goes live.
Expand Down
2 changes: 1 addition & 1 deletion contents/docs/semantic-layer/start-here.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,7 @@ One conversation gave you the full trust chain:

From here:

- [`setting-up-data-catalog` skill](https://github.com/PostHog/posthog/blob/master/products/data_catalog/skills/setting-up-data-catalog/SKILL.md) – with the [PostHog AI plugin](https://github.com/PostHog/ai-plugin) installed, ask your agent to set up the data catalog, and it proposes metrics from your project's most-used insights for you to review
- [`setting-up-data-catalog` skill](https://github.com/PostHog/posthog/blob/master/products/data_catalog/skills/setting-up-data-catalog/SKILL.md) – with the [PostHog MCP server](/docs/model-context-protocol) connected, ask your agent to set up the data catalog, and it proposes metrics from your project's most-used insights for you to review
- [Metrics](/docs/semantic-layer/metrics) – definition kinds, the lifecycle, and what happens when a metric drifts
- [Certifications and relationships](/docs/semantic-layer/certifications-and-relationships) – govern the tables your answers come from, not just the answers
- [MCP tools](/docs/semantic-layer/mcp-tools) – the full tool reference and more example prompts
8 changes: 3 additions & 5 deletions contents/docs/skills/index.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -6,9 +6,9 @@ Skills are versioned, reusable instructions for AI agents, stored centrally in P

Skills also power the rest of self-driving PostHog: every [scout](/docs/self-driving/scouts) is a skill you can read, edit, or write yourself. And you don't have to start from scratch – browse the [community skills store](/docs/skills/community) to install skills shared by other PostHog users, or publish your own.

> **Shortcut: install the PostHog AI plugin.**
> **Shortcut: connect the PostHog MCP server.**
>
> The [PostHog AI plugin](https://github.com/PostHog/ai-plugin) ships an official [`skills-store` skill](https://github.com/PostHog/ai-plugin/tree/main/skills/skills-store) – maintained alongside the [canonical version in the PostHog monorepo](https://github.com/PostHog/posthog/tree/master/products/skills/skills/skills-store) – that teaches your agent exactly how to discover, load, create, and update skills via the PostHog MCP tools. Install the plugin in [Claude Code](/docs/model-context-protocol/claude-code), [Codex](/docs/model-context-protocol/codex), Cursor, or Gemini CLI and your agent already knows how to use the store.
> PostHog publishes an official [`skills-store` skill](https://github.com/PostHog/posthog/tree/master/products/skills/skills/skills-store) that teaches your agent exactly how to discover, load, create, and update skills via the PostHog MCP tools. It ships with the [PostHog MCP server](/docs/model-context-protocol), so once your agent is connected it can load the skill when it needs it – there's nothing extra to install.

## Why use PostHog as your skills store?

Expand Down Expand Up @@ -188,9 +188,7 @@ This works identically in Claude Code, Cursor, Codex, or any other MCP-connected

## Step 3: Set up a bridge skill (optional shortcut)

Step 2 is all you need. If you've installed the [PostHog AI plugin](https://github.com/PostHog/ai-plugin), you can skip this step entirely – the plugin's `skills-store` skill already provides the bridge.

Otherwise, if your agent supports local skills (like Claude Code's `.claude/skills/` directory), you can drop in a tiny DIY bridge that bootstraps into the PostHog skills store. You only need **one** generic bridge:
Step 2 is all you need. If your agent supports local skills (like Claude Code's `.claude/skills/` directory), you can drop in a tiny DIY bridge that bootstraps into the PostHog skills store. You only need **one** generic bridge:

### Claude Code example

Expand Down
6 changes: 3 additions & 3 deletions src/components/AgentSkillsList/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -10,9 +10,9 @@ via the `posthog-main-repo` `gatsby-source-git` source – see `gatsby/onCreateN
`useAgentSkills()` in `src/hooks/skills.tsx`. The list updates automatically on every site build;
nothing here is hand-maintained.

Installation guidance lives in the docs page prose and points at the
[PostHog AI plugin](https://github.com/PostHog/ai-plugin), whose README is the canonical,
always-current install reference – deliberately not duplicated here or in the page.
Installation guidance lives in the docs page prose: the skills ship with the
[PostHog MCP server](/docs/model-context-protocol), so there is nothing to install beyond
connecting it – deliberately not duplicated here.

## Usage

Expand Down
2 changes: 1 addition & 1 deletion src/components/MCPPage/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,7 @@ To keep the brief true:
- **Quotes** (`MCPNoUI`): each is the `mcp` product quote (`quotes.<author>.products.mcp`) of a person in `mcpQuotes`, from `useCustomers`. Use only a real quote from a published customer story, word for word. Mark a cut with `[…]`. The first quote gets the big card, so put the longest one first.
- **Stat** (`mcpStat`): use only a number from a published post. The current one is from [How AI agents behave](/blog/how-ai-agents-behave).
- **Chips** (`whatItIs` in `WhatIsMCP`): one plain claim each, in two lines or fewer. "Light on context" is true in CLI mode (the default in Claude, Claude Code, Cowork, and Codex), not in tools mode (the default in Cursor).
- **Clients** (`connectors`): check the list against the [MCP overview](/docs/model-context-protocol) before you change it. The "Want skills too?" note describes the [PostHog AI plugin](https://github.com/PostHog/ai-plugin).
- **Clients** (`connectors`): check the list against the [MCP overview](/docs/model-context-protocol) before you change it. The "Want skills too?" note says skills ship with the MCP server and links to [Skills](/docs/skills).
- **FAQ** (`faq`): each answer is a short version of an [MCP FAQ](/docs/model-context-protocol/faq) answer, and links to that heading. Check the docs before you change a number. Update the link anchor when a docs heading changes.
- **Setup CTA** (`MCPSetupCTA`): the copy comes from "Install PostHog from chat" in the [MCP use cases](/docs/model-context-protocol/use-cases).
- **Header:** the copy comes from the [MCP docs](/docs/model-context-protocol). Under `lg`, the install card is hidden and a "Read the setup guide" button shows, because a phone visitor is unlikely to run an install command.
Expand Down
2 changes: 1 addition & 1 deletion src/components/MCPPage/agentBrief.ts
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,7 @@ Any client that speaks MCP works. Setup steps for each:
- Zed: https://posthog.com/docs/model-context-protocol/zed.md
- Other clients: https://posthog.com/docs/model-context-protocol.md

Plugin: the PostHog AI plugin adds the MCP server and PostHog skills to Claude Code, Codex, Cursor, and Gemini CLI: https://github.com/PostHog/ai-plugin
Skills: PostHog skills ship with the MCP server, so a connected agent loads the one it needs when a task calls for it. There's nothing extra to install: https://posthog.com/docs/skills.md

Platforms that connect to PostHog through the MCP server:

Expand Down
11 changes: 5 additions & 6 deletions src/components/MCPPage/index.tsx
Original file line number Diff line number Diff line change
Expand Up @@ -332,13 +332,12 @@ function MCPConnectors(): JSX.Element {
<div className="mt-6 flex items-start gap-3">
<IconPlug className="mt-0.5 size-5 shrink-0 text-blue" />
<p className="m-0 text-sm leading-relaxed text-secondary">
<span className="font-semibold text-primary">Want skills too?</span> The{' '}
<Link to="https://github.com/PostHog/ai-plugin" className="font-semibold underline">
PostHog AI plugin
<span className="font-semibold text-primary">Want skills too?</span> PostHog{' '}
<Link to="/docs/skills" className="font-semibold underline">
skills
</Link>{' '}
installs the same MCP server plus more than 30 PostHog skills in one step. It works in Claude Code,
Codex, Cursor, and Gemini CLI. If you connect the server by hand, you get the same tools without the
bundled skills.
ship with the MCP server, so there's nothing extra to install. Once your agent is connected, it
loads the skill it needs when a task calls for it.
</p>
</div>
</section>
Expand Down
4 changes: 2 additions & 2 deletions src/components/WebMCP/index.tsx
Original file line number Diff line number Diff line change
Expand Up @@ -55,9 +55,9 @@ const SKILL_RAW_BASE = 'https://raw.githubusercontent.com/PostHog/posthog/master
const SKILL_SOURCE_BASE = 'https://github.com/PostHog/posthog/tree/master'

// The two install paths the docs give people, see contents/docs/ai-observability/skills.mdx.
const SKILLS_PLUGIN_URL = 'https://github.com/PostHog/ai-plugin'
const SKILLS_MCP_URL = 'https://posthog.com/docs/model-context-protocol'
const SKILLS_RELEASE_URL = 'https://github.com/PostHog/posthog/releases/tag/agent-skills-latest'
const SKILLS_INSTALL_LINE = `Install: the PostHog AI plugin (${SKILLS_PLUGIN_URL}) bundles every PostHog skill for Claude Code, Codex, Cursor, and Gemini CLI. Or download skills.zip from ${SKILLS_RELEASE_URL} and copy a skill folder into the agent's skills directory, for example .claude/skills/.`
const SKILLS_INSTALL_LINE = `Install: every PostHog skill ships with the PostHog MCP server (${SKILLS_MCP_URL}), so a connected agent loads the one it needs when a task calls for it. Or download skills.zip from ${SKILLS_RELEASE_URL} and copy a skill folder into the agent's skills directory, for example .claude/skills/.`

const NO_INPUT_SCHEMA = { type: 'object', properties: {} }

Expand Down
Loading