@nanobpm/nano-coder 0.62.3 → 0.63.1

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 +13 -2
  2. package/package.json +5 -5
package/README.md CHANGED
@@ -117,7 +117,9 @@ A steer that arrives too late to join the turn runs as an ordinary prompt, or is
117
117
  answered with `stopReason: "cancelled"` if the turn was cancelled. Slash-command prompts
118
118
  and other requests that arrive mid-turn are handled after the turn ends. Everything
119
119
  handled after the turn, late steers included, runs in the order the client sent it.
120
- `stopReason` is `end_turn`, `cancelled`, or `max_turn_requests`.
120
+ `stopReason` is `end_turn`, `cancelled`, or `max_turn_requests`. When the provider reported
121
+ token counts, the result also carries ACP's `usage`: `{"inputTokens", "outputTokens", "totalTokens"}`,
122
+ summed over the turn's model calls.
121
123
 
122
124
  **Streaming updates.** During a turn the harness sends `session/update` notifications:
123
125
  `agent_message_chunk` (with a `messageId`), `tool_call` (`toolCallId`, `title` = tool name,
@@ -680,7 +682,7 @@ kind = "anthropic"
680
682
  base_url = "https://llm-gateway.example.com/anthropic/v1"
681
683
  api_key_env = "WORK_GATEWAY_KEY" # or api_key = "..." (prefer the env var)
682
684
  # api_key_command = "op read op://vault/gateway/key" # used when the env var is unset
683
- timeout_secs = 300 # idle timeout: max silence between streamed bytes (default 600)
685
+ timeout_secs = 300 # idle timeout: max silence between streamed bytes (default 120, or $NANO_CODER_PROVIDER_TIMEOUT_SECS)
684
686
  max_retries = 3
685
687
  ```
686
688
 
@@ -886,6 +888,15 @@ Other per-provider fields: `replay_reasoning`, `max_tokens_param` (`max_tokens`,
886
888
  which is the `openai` default), `retry_initial_backoff_ms`, `retry_max_backoff_ms`,
887
889
  `retryable_statuses` and `offline_budget_ms`.
888
890
 
891
+ `timeout_secs` is an *idle* timeout: it resets on every received byte, so a slow but steady
892
+ stream is never cut off — only a genuinely stalled connection trips it. The default is 120 s
893
+ (`NANO_CODER_PROVIDER_TIMEOUT_SECS` overrides the default; an explicit `timeout_secs` wins
894
+ over both). When nano-coder runs under a supervisor that idle-kills a silent run — e.g. the
895
+ c8ctl-nano worker, which kills a run after 300 s of no output — `timeout_secs` must stay
896
+ **below** that budget: at or above it the worker kills the whole run before a stalled stream
897
+ is ever detected and retried. A supervising worker can pass its own derived budget (e.g. its
898
+ idle-kill budget / 2) as `NANO_CODER_PROVIDER_TIMEOUT_SECS` so the two timeouts cannot drift.
899
+
889
900
  A transport failure with no HTTP response (DNS, connection refused or reset, connect or
890
901
  idle timeout) is treated as the network being down rather than a failed attempt: nano-coder
891
902
  backs off (`retry_initial_backoff_ms` doubling to `retry_max_backoff_ms`) and keeps waiting,
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@nanobpm/nano-coder",
3
- "version": "0.62.3",
3
+ "version": "0.63.1",
4
4
  "description": "A 6MB coding agent. Run a fleet on your laptop.",
5
5
  "license": "Apache-2.0",
6
6
  "repository": {
@@ -27,9 +27,9 @@
27
27
  "node": ">=18"
28
28
  },
29
29
  "optionalDependencies": {
30
- "@nanobpm/nano-coder-darwin-arm64": "0.62.3",
31
- "@nanobpm/nano-coder-darwin-x64": "0.62.3",
32
- "@nanobpm/nano-coder-linux-arm64": "0.62.3",
33
- "@nanobpm/nano-coder-linux-x64": "0.62.3"
30
+ "@nanobpm/nano-coder-darwin-arm64": "0.63.1",
31
+ "@nanobpm/nano-coder-darwin-x64": "0.63.1",
32
+ "@nanobpm/nano-coder-linux-arm64": "0.63.1",
33
+ "@nanobpm/nano-coder-linux-x64": "0.63.1"
34
34
  }
35
35
  }