Skip to content

Build AI methods with your coding agent, then run them anywhere: as an MCP for chatbots, as a webapp for people, or via API for your software.

fix: update Makefile cleanenv description and enhance reasoning controls with xhigh level

Louis Choquelbb8b169d36a4
diff --git a/Makefile b/Makefile
index 5588f38b2..a9b9dde1b 100644
--- a/Makefile
+++ b/Makefile
@@ -87,7 +87,7 @@ make rtm - Shorthand -> regenerate-test-models
make insert-skeleton - Insert skeleton from $(SKELETON_DIR)
make up - Shorthand -> generate-mthds-schema update-gateway-models up-kit-configs rules
-make cleanenv - Remove virtual env and lock files
+make cleanenv - Remove virtual env
make cleanderived - Remove extraneous compiled files, caches, logs, etc.
make cleanall - Remove all -> cleanenv + cleanderived
diff --git a/docs/features/distributed-execution.md b/docs/features/distributed-execution.md
index afb04aba7..b221736dc 100644
--- a/docs/features/distributed-execution.md
+++ b/docs/features/distributed-execution.md
@@ -9,7 +9,7 @@ Run your `.mthds` methods as durable Temporal workflows.
## Overview
-Pipelex pipelines normally run in-process. With the optional `pipelex[temporal]` integration, the same pipelines run as Temporal workflows: each pipe becomes a workflow, child pipes become child workflows, and every LLM call, image generation, or document extraction becomes an activity. Temporal handles durability, retries, scheduling, and visibility; Pipelex handles the AI work. Flip `[temporal] is_enabled = true` in `pipelex.toml` and the same methods run distributed without changing a line of method code.
+Pipelex pipelines normally run in-process. With the optional `pipelex[temporal]` integration, the same pipelines run as Temporal workflows: each pipe becomes a workflow, child pipes become child workflows, and every LLM call, image generation, or document extraction becomes an activity. Temporal handles durability, retries, scheduling, and visibility; Pipelex handles the AI work. Flip `[temporal] is_enabled = true` in `.pipelex/pipelex.toml` and the same methods run distributed without changing a line of method code.
## Supported deployment patterns
diff --git a/docs/under-the-hood/reasoning-controls.md b/docs/under-the-hood/reasoning-controls.md
index 1be05673c..e15b181e5 100644
--- a/docs/under-the-hood/reasoning-controls.md
+++ b/docs/under-the-hood/reasoning-controls.md
@@ -178,6 +178,7 @@ minimal = "minimal"
low = "low"
medium = "medium"
high = "high"
+xhigh = "xhigh"
max = "xhigh"
```
@@ -188,6 +189,7 @@ max = "xhigh"
| `LOW` | `"low"` |
| `MEDIUM` | `"medium"` |
| `HIGH` | `"high"` |
+| `XHIGH` | `"xhigh"` |
| `MAX` | `"xhigh"` |
!!! note
@@ -208,6 +210,7 @@ minimal = "low"
low = "low"
medium = "medium"
high = "high"
+xhigh = "xhigh"
max = "max"
```
@@ -218,6 +221,7 @@ max = "max"
| `LOW` | `"low"` |
| `MEDIUM` | `"medium"` |
| `HIGH` | `"high"` |
+| `XHIGH` | `"xhigh"` |
| `MAX` | `"max"` |
Both modes first check `anthropic_config.effort_to_level_map` to gate reasoning. If the map returns `"disabled"` (e.g., for `NONE` effort), thinking is disabled entirely — no `thinking` parameter is sent to the SDK.
@@ -244,6 +248,7 @@ minimal = "low"
low = "low"
medium = "medium"
high = "high"
+xhigh = "high"
max = "high"
```
@@ -266,6 +271,7 @@ If the level map returns `"disabled"` (e.g., for `NONE` effort), thinking is dis
| `LOW` | `1024` |
| `MEDIUM` | `5000` |
| `HIGH` | `16384` |
+| `XHIGH` | `32768` |
| `MAX` | `65536` |
**`reasoning_budget`** (explicit) passes through directly as `thinking_budget`. When `max_tokens` is known, the budget is capped to `min(budget, max_tokens - 1)`.
@@ -289,6 +295,7 @@ minimal = "reasoning"
low = "reasoning"
medium = "reasoning"
high = "reasoning"
+xhigh = "reasoning"
max = "reasoning"
```
@@ -340,6 +347,7 @@ minimal = 512
low = 1024
medium = 5000
high = 16384
+xhigh = 32768
max = 65536
[cogt.llm_config.effort_to_budget_maps.gemini]
@@ -348,6 +356,7 @@ minimal = 512
low = 1024
medium = 5000
high = 16384
+xhigh = 32768
max = 65536
```