Pi - The Context Company

Set TCC environment variables

The Pi extension and SDK both default to using the TCC_API_KEY environment variable.

.env.local

TCC_API_KEY="your-api-key"

Instrument Pi

Use the SDK path when you create Pi sessions or event streams from your own TypeScript code. Use the Extension path when you run Pi from the command line.

SDK
Extension

Step 1: Install dependencies

pnpm add @contextcompany/pi @mariozechner/pi-coding-agent

Step 2: Add instrumentation to your agent

Import the instrumentPiSession function and call it with your Pi agent session. This subscribes to all agent events and automatically exports traces to The Context Company.

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";

const { session } = await createAgentSession();

instrumentPiSession(session);

await session.prompt("What files are in the current directory?");

That’s it! All your agent runs will now be automatically tracked in The Context Company.

Step 3: Instrument Pi event streams

If you’re working with Pi event streams directly instead of a session object, you can use instrumentPiEventStream to wrap an AsyncIterable of events. This forwards each event through the telemetry listener and automatically exports traces to The Context Company.

agent.ts

import { instrumentPiEventStream } from "@contextcompany/pi";

// Wrap your event stream with instrumentation
const instrumentedStream = instrumentPiEventStream(events);

for await (const event of instrumentedStream) {
  // Process events as usual — telemetry is captured automatically
  console.log(event);
}

The instrumentPiEventStream function accepts the same config options as instrumentPiSession, including metadata, sessionId, conversational, and debug.

agent.ts

import { instrumentPiEventStream } from "@contextcompany/pi";

const instrumentedStream = instrumentPiEventStream(events, {
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
  },
  sessionId: "some-session-id",
  conversational: true,
});

for await (const event of instrumentedStream) {
  console.log(event);
}

You can retrieve the last run ID from the instrumented stream:

agent.ts

const instrumentedStream = instrumentPiEventStream(events);

for await (const event of instrumentedStream) {
  // ...
}

const lastRunId = instrumentedStream.getLastRunId();

Step 1: Install the Pi CLI extension

If you use Pi from the command line, install The Context Company as a Pi extension. The extension automatically records Pi agent runs, messages, and tool executions without changing your agent code.

pi install npm:@contextcompany/pi

Step 2: Start Pi

pi

Run /tcc-status inside Pi to confirm that telemetry is active. If a run has completed, the command also returns the last TCC run ID recorded by the extension.

Step 3: Add extension metadata

To attach metadata to runs captured by the extension, add a .pi/tcc.json file to your project.

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.sessionId": "conversation-123",
    "tcc.conversational": true
  }
}

You can also set default metadata globally in ~/.pi/agent/tcc.json. Project metadata in .pi/tcc.json is merged on top of global metadata.

The extension uses TCC_API_KEY for authentication. Debug logging is covered in Debug mode.

The extension handles Pi CLI runs automatically. SDK-only controls such as instrumentPiEventStream, unsubscribe, and assigning a run ID before each prompt with setRunId() are only available when you instrument Pi from code.

Adding custom metadata

Custom metadata allows you to add additional properties to your agent runs. This is particularly useful for tying agent runs to your own specific business logic, letting you filter and analyze agent runs by user, organization, feature, or some other dimension.

SDK
Extension

Pass custom metadata and reserved tcc.* metadata as key-value pairs to the metadata object in the config.

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";

const { session } = await createAgentSession();

instrumentPiSession(session, {
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
    "tcc.sessionId": "conversation-123",
    "tcc.conversational": true,
    "tcc.agent": "coding-agent",
    "tcc.userId": "user-123",
    "tcc.orgId": "org-456",
  },
});

Add the same metadata fields to .pi/tcc.json:

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.sessionId": "conversation-123",
    "tcc.conversational": true,
    "tcc.agent": "coding-agent",
    "tcc.userId": "user-123",
    "tcc.orgId": "org-456"
  }
}

Agent runs are automatically indexed by your custom metadata fields and can be filtered directly in the dashboard.

The tcc.* namespace is reserved. Only the reserved TCC metadata keys (tcc.runId, tcc.sessionId, tcc.conversational, tcc.agent, tcc.userId, tcc.userName, tcc.orgId, tcc.orgName) are recognized; any other tcc.* keys are ignored. None of them appear in your custom metadata.A Pi session can fire many runs, but metadata is captured once and reused for every run in that session. Do not put tcc.runId here. In SDK instrumentation, use instrumentation.setRunId(...) before each prompt to assign a per-run ID.

Adding user feedback

User feedback allows you to collect score (thumbs up & thumbs down) and text feedback (up to 2000 characters) from end users on your agent runs. This is useful for tracking user satisfaction, identifying problematic responses, and filtering agent runs in the dashboard to focus on positive or negative feedback.

Step 1: Generate and pass a run ID

Use setRunId() on the instrumentation object to assign a unique run ID before each prompt. After the run completes, retrieve it with getLastRunId().

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";
import { randomUUID } from "crypto";

const { session } = await createAgentSession();

const instrumentation = instrumentPiSession(session);

// Generate a unique run ID (must be a UUID) before each prompt
const runId = randomUUID();
instrumentation.setRunId(runId);

await session.prompt("What files are in the current directory?");

// Return the runId to your client
return { runId };

Programmatic feedback is only supported with SDK instrumentation. The Pi CLI extension can show the last completed run ID with /tcc-status for debugging, but it does not provide a stable app-facing setRunId() flow for collecting feedback from users.

Step 2: Submit feedback from your client

Store the runId on your client, then when the user provides feedback, submit it using the submitFeedback function. Both score and text are optional individually, but each request must include at least one of them: score is the thumbs rating. Use only "thumbs_up" or "thumbs_down". text is written feedback from your user, up to 2000 characters.

feedback.ts

import { submitFeedback } from "@contextcompany/pi";

// Submit score and/or text feedback:
await submitFeedback({
  runId: runId, // The run ID from your agent execution
  score: "thumbs_up", // Optional thumbs rating: "thumbs_up" or "thumbs_down"
  text: "This was a helpful response!", // Optional written user feedback, up to 2000 characters
});

Agent runs with feedback can be filtered in the dashboard using the feedback filter.

Tracking agent sessions

Agent sessions represent multiple agent runs that are grouped together. The most common use case is tracking entire conversations between a human user and an AI agent in chatbot interfaces. Agent sessions can be tracked by setting a session ID.

SDK
Extension

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";

const { session } = await createAgentSession();

instrumentPiSession(session, {
  sessionId: "conversation-123", // Track agent sessions
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
  },
});

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.sessionId": "conversation-123"
  }
}

The value of sessionId should be a unique identifier for the agent session. This can be any string, but it’s generally recommended to use a UUID. Agent sessions are automatically indexed and can be filtered directly in the dashboard.

Marking runs as conversational

A conversational run is an agent run that was initiated by a user. Marking a run as conversational tells The Context Company that this run involves direct user interaction. This is important because conversational runs are the only runs monitored for user insights, such as user confusion, frustration, or any other custom insights you want to track. Runs that are not marked as conversational (e.g. background jobs, cron tasks, or internal automations) are excluded from user insight analysis. Mark a run as conversational by setting tcc.conversational in extension metadata or conversational in SDK config.

SDK
Extension

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";

const { session } = await createAgentSession();

instrumentPiSession(session, {
  conversational: true,
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
  },
});

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.conversational": true
  }
}

The conversational, sessionId, and runId options on instrumentPiSession are typed shortcuts for the reserved TCC metadata keys.

Identifying the agent

If your product ships more than one named agent, set the reserved tcc.agent metadata key to scope the session to a specific agent. The dashboard’s top-level agent selector, per-agent patterns and recaps, and the agent filter on the REST API and MCP tools all read from this key.

SDK
Extension

agent.ts

import { instrumentPiSession } from "@contextcompany/pi";

instrumentPiSession(session, {
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
    "tcc.agent": "coding-agent",
  },
});

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.agent": "coding-agent"
  }
}

Agent names that collide with reserved dashboard routes (for example runs, sessions, patterns, recaps, overview, search, failures, feedback, tools, topics, views, settings, mcp-and-api) are dropped.

Identifying users and organizations

Attach the end user and their organization to a Pi session as first-class identity using the reserved tcc.userId, tcc.userName, tcc.orgId, and tcc.orgName metadata keys. This is not the same as adding a userId field to custom metadata — these keys promote user and org identity to dedicated dashboard filters and unlock native user/org search, per-user views, and per-org analytics. See User and organization identity for the full concept. Set these whenever you have a stable identifier for the end user or their organization in your product.

SDK
Extension

agent.ts

import { instrumentPiSession } from "@contextcompany/pi";

instrumentPiSession(session, {
  metadata: {
    environment: "staging",
    feature: "repo-review",
    customerTier: "enterprise",
    "tcc.userId": "user-123",
    "tcc.userName": "Jane Doe",
    "tcc.orgId": "org-456",
    "tcc.orgName": "Acme Inc.",
  },
});

.pi/tcc.json

{
  "metadata": {
    "environment": "staging",
    "feature": "repo-review",
    "customerTier": "enterprise",
    "tcc.userId": "user-123",
    "tcc.userName": "Jane Doe",
    "tcc.orgId": "org-456",
    "tcc.orgName": "Acme Inc."
  }
}

tcc.userName and tcc.orgName require the corresponding ID (tcc.userId / tcc.orgId) to also be set. Names without IDs are dropped.

Cleanup

The instrumentPiSession function returns an object with an unsubscribe method you can call to stop listening for events. This is only needed for SDK instrumentation; the Pi CLI extension manages its own lifecycle.

agent.ts

const instrumentation = instrumentPiSession(session);

// Later, when you're done:
instrumentation.unsubscribe();

Debug mode

You can enable debug mode, which will log any events that are captured and sent. For the Pi CLI extension, set TCC_DEBUG=true before starting Pi:

TCC_DEBUG=true pi

For SDK instrumentation, pass debug: true:

agent.ts

import { createAgentSession } from "@mariozechner/pi-coding-agent";
import { instrumentPiSession } from "@contextcompany/pi";

const { session } = await createAgentSession();

instrumentPiSession(session, {
  debug: true, // Enable debug mode
});

​Set TCC environment variables

​Instrument Pi

​Step 1: Install dependencies

​Step 2: Add instrumentation to your agent

​Step 3: Instrument Pi event streams

​Step 1: Install the Pi CLI extension

​Step 2: Start Pi

​Step 3: Add extension metadata

​Adding custom metadata

​Adding user feedback

​Step 1: Generate and pass a run ID

​Step 2: Submit feedback from your client

​Tracking agent sessions

​Marking runs as conversational

​Identifying the agent

​Identifying users and organizations

​Cleanup

​Debug mode

Set TCC environment variables

Instrument Pi

Step 1: Install dependencies

Step 2: Add instrumentation to your agent

Step 3: Instrument Pi event streams

Step 1: Install the Pi CLI extension

Step 2: Start Pi

Step 3: Add extension metadata

Adding custom metadata

Adding user feedback

Step 1: Generate and pass a run ID

Step 2: Submit feedback from your client

Tracking agent sessions

Marking runs as conversational

Identifying the agent

Identifying users and organizations

Cleanup

Debug mode