@cat-factory/observability-langfuse 0.9.45 → 0.9.47

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +3 -3
  2. package/package.json +2 -2
package/README.md CHANGED
@@ -3,9 +3,9 @@
3
3
  Opt-in [Langfuse](https://langfuse.com) trace sink for the Agent Architecture Board.
4
4
 
5
5
  It implements the runtime-neutral `LlmTraceSink` port from `@cat-factory/kernel`, so
6
- when wired into a facade every LLM call container-agent calls (through the LLM proxy)
6
+ when wired into a facade every LLM call: container-agent calls (through the LLM proxy)
7
7
  **and** inline calls (requirements review, document planner, fragment selector, inline
8
- agent) surfaces in Langfuse as a generation grouped under its run's trace, plus
8
+ agent): surfaces in Langfuse as a generation grouped under its run's trace, plus
9
9
  optional container tool spans.
10
10
 
11
11
  ## Why fetch-only
@@ -13,7 +13,7 @@ optional container tool spans.
13
13
  The sink talks to Langfuse's public **ingestion API** (`POST /api/public/ingestion`,
14
14
  HTTP Basic auth, batched JSON events) using only the global `fetch` / `crypto` /
15
15
  `btoa`. It deliberately does **not** depend on the official `langfuse` Node SDK or any
16
- `@opentelemetry/*` package those rely on Node-only APIs that are unavailable on the
16
+ `@opentelemetry/*` package: those rely on Node-only APIs that are unavailable on the
17
17
  Cloudflare Worker runtime (workerd). Using the raw ingestion API keeps the sink
18
18
  byte-for-byte identical on both the Worker and Node facades.
19
19
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@cat-factory/observability-langfuse",
3
- "version": "0.9.45",
3
+ "version": "0.9.47",
4
4
  "description": "Opt-in Langfuse trace sink for the Agent Architecture Board. A fetch-based LlmTraceSink that streams LLM generations (and container tool spans) to Langfuse's ingestion API — runs unchanged on both the Cloudflare Worker (workerd) and Node facades.",
5
5
  "repository": {
6
6
  "type": "git",
@@ -24,7 +24,7 @@
24
24
  "access": "public"
25
25
  },
26
26
  "dependencies": {
27
- "@cat-factory/kernel": "0.217.0"
27
+ "@cat-factory/kernel": "0.219.0"
28
28
  },
29
29
  "devDependencies": {
30
30
  "typescript": "7.0.2",