You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
AgentPond is a lightweight trace backend and CLI for AI applications. It keeps raw traces in object storage you control and syncs them into a local DuckDB cache for fast analysis by your coding agent. Use it with Firebase Storage, Supabase Storage, Amazon S3, Google Cloud Storage, Vercel Blob, or custom infrastructure. Firebase, Supabase, and Vercel projects include automated setup; other deployments use a manual setup path.
17
+
AgentPond is a lightweight trace backend and CLI for AI applications. It keeps raw traces in object storage you control and syncs them into a local DuckDB cache for fast analysis by your coding agent. It supports all bucket providers from [Files SDK](https://github.com/haydenbleasel/files-sdk) such as Amazon S3, Google Cloud Storage, R2, and MinIO; For platforms like Firebase, Supabase, and Vercel there are explicit one-command quick starts.
18
18
19
19
## How it works
20
20
@@ -24,21 +24,15 @@ Object storage is the durable source of truth. The local DuckDB database is a re
24
24
25
25
## Getting started
26
26
27
-
Start with the [Manual deployment setup](./docs/getting-started/manual-setup.md) to configure object storage, instrument the application, and sync its traces into AgentPond.
27
+
Start with the [Manual deployment setup](./docs/getting-started/manual-setup.md) to configure object storage, instrument your application, and sync its traces into AgentPond.
28
28
29
-
For Firebase, Supabase, and Vercel projects, AgentPond provides automated quick starts:
All require Node.js 22 or newer for the CLI. From your project folder, run:
29
+
For [Firebase](./docs/getting-started/firebase.md), [Supabase](./docs/getting-started/supabase.md), and [Vercel](./docs/getting-started/vercel.md) projects, to get started, just run:
36
30
37
31
```sh
38
32
npx agentpond init
39
33
```
40
34
41
-
The command detects the platform, installs AgentPond's instrumentation and analytics skills, and prints a install prompt for your coding agent. Your agent inspects then the application, proposes a setup, implements it after confirmation, and verifies a real trace end to end.
35
+
The command installs AgentPond's instrumentation and analytics skills, and prints a install prompt for your coding agent. Your coding agent inspects then the application, proposes a setup, implements it after confirmation, and verifies a real trace end to end.
42
36
43
37
Once the application has emitted a trace:
44
38
@@ -47,26 +41,9 @@ npx agentpond sync
47
41
npx agentpond traces list --limit 10
48
42
```
49
43
50
-
## Analyze traces
51
-
52
-
Use focused commands for individual traces and sessions:
53
-
54
-
```sh
55
-
npx agentpond traces get <trace-id>
56
-
npx agentpond observations list --traceId <trace-id>
57
-
npx agentpond sessions get <session-id>
58
-
npx agentpond scores list --traceId <trace-id>
59
-
```
60
-
61
-
Use DuckDB SQL for aggregation, joins, time windows, or cost analysis:
62
-
63
-
```sh
64
-
npx agentpond sql "select id, name, session_id, total_cost from traces order by start_time desc limit 10"
65
-
```
66
-
67
44
## Features
68
45
69
-
- Direct OpenTelemetry export to Firebase Storage, Supabase Storage, S3, GCS, and Vercel Blob
46
+
- Direct OpenTelemetry export through Files SDK, Firebase Storage, Supabase Storage, and Vercel Blob
70
47
- Langfuse-compatible and OTLP HTTP ingestion adapters
71
48
- Incremental object-store synchronization
72
49
- Local DuckDB projections for traces, observations, sessions, and scores
0 commit comments