fix: default max_rounds to 0 (unlimited) (#540)

This commit is contained in:
tlongwell-block
2026-05-11 20:09:25 +00:00
committed by GitHub
parent 1b9f6169ee
commit d953d0ecb4
2 changed files with 3 additions and 3 deletions
+2 -2
View File
@@ -133,7 +133,7 @@ Everything is environment variables. No flags, no config files. (We are a subpro
| `OPENAI_COMPAT_BASE_URL` | `https://api.openai.com/v1` | Point at vLLM, llama.cpp, OpenRouter, Ollama, etc. |
| `SPROUT_AGENT_SYSTEM_PROMPT` | built-in | Inline system prompt. |
| `SPROUT_AGENT_SYSTEM_PROMPT_FILE` | — | File path. Mutually exclusive with the above. |
| `SPROUT_AGENT_MAX_ROUNDS` | `16` | Tool-loop iteration cap. |
| `SPROUT_AGENT_MAX_ROUNDS` | `0` | Tool-loop iteration cap. 0 = unlimited. |
| `SPROUT_AGENT_MAX_OUTPUT_TOKENS` | `4096` | Per LLM call. |
| `SPROUT_AGENT_LLM_TIMEOUT_SECS` | `120` | |
| `SPROUT_AGENT_TOOL_TIMEOUT_SECS` | `660` | Per-tool call timeout in seconds |
@@ -223,7 +223,7 @@ The trust boundary is **the operator who launched the agent**. The harness, MCP
| Tool description bytes | 1 KiB | `MAX_DESCRIPTION_BYTES` |
| Tool schema bytes | 4 KiB | `MAX_SCHEMA_BYTES` (oversize → replaced with `{}`) |
| Tool calls per turn | 64 | `MAX_TOOL_CALLS_PER_TURN` |
| Loop rounds | 16 | `SPROUT_AGENT_MAX_ROUNDS` |
| Loop rounds | 0 (unlimited) | `SPROUT_AGENT_MAX_ROUNDS` |
| LLM call timeout | 120 s | `SPROUT_AGENT_LLM_TIMEOUT_SECS` |
| Tool call timeout | 660 s | `SPROUT_AGENT_TOOL_TIMEOUT_SECS` |
+1 -1
View File
@@ -92,7 +92,7 @@ impl Config {
model,
base_url,
anthropic_api_version: env_or("ANTHROPIC_API_VERSION", "2023-06-01"),
max_rounds: parse_env("SPROUT_AGENT_MAX_ROUNDS", 16)?,
max_rounds: parse_env("SPROUT_AGENT_MAX_ROUNDS", 0)?,
max_output_tokens: parse_env("SPROUT_AGENT_MAX_OUTPUT_TOKENS", 4096)?,
llm_timeout: Duration::from_secs(parse_env("SPROUT_AGENT_LLM_TIMEOUT_SECS", 120)?),
tool_timeout: Duration::from_secs(parse_env("SPROUT_AGENT_TOOL_TIMEOUT_SECS", 660)?),