@evolvingmachines/evolve 0.0.56 → 0.0.57

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.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@evolvingmachines/evolve",
3
- "version": "0.0.56",
3
+ "version": "0.0.57",
4
4
  "keywords": [
5
5
  "ai",
6
6
  "agents",
@@ -129,9 +129,9 @@
129
129
  },
130
130
  "dependencies": {
131
131
  "@agentclientprotocol/sdk": "^0.5.1",
132
- "@evolvingmachines/daytona": "0.0.56",
133
- "@evolvingmachines/e2b": "0.0.56",
134
- "@evolvingmachines/modal": "0.0.56",
132
+ "@evolvingmachines/daytona": "0.0.57",
133
+ "@evolvingmachines/e2b": "0.0.57",
134
+ "@evolvingmachines/modal": "0.0.57",
135
135
  "ajv": "^8.17.1",
136
136
  "p-map": "^7.0.2",
137
137
  "smol-toml": "^1.7.0",
@@ -14,7 +14,7 @@ Python SDK: `pip install evolvingmachines-evolve` (`import evolve`). The manual
14
14
 
15
15
  ## Start here
16
16
 
17
- This file is a pointer, not the manual. The manual ships inside the CLI and always matches the installed version. Before running any `evolve` command, load it:
17
+ This file is a pointer, not the manual. The manual ships inside the CLI and always matches the installed version. Keep it current: `npm i -g @evolvingmachines/evolve@latest`, then `evolve skills install --force` refreshes this file. Before running any `evolve` command, load it:
18
18
 
19
19
  ```bash
20
20
  evolve skills get evals # the index of the documentation: every page, one line each
@@ -16,6 +16,14 @@ This page takes you from an empty terminal to a downloaded trial: one API key, a
16
16
  Create a key on the dashboard's [API keys page](https://dashboard.evolvingmachines.ai/api-keys), then export it. See [installation](/getting-started/installation) for the SDKs.
17
17
  </Step>
18
18
 
19
+ <Step title="Give your coding agent the manual">
20
+ ```bash
21
+ evolve skills install
22
+ ```
23
+
24
+ Your agent then reads these docs with `evolve skills get evals`, and the `create-task`, `rewardkit`, `create-adapter` and `publish` skills by name. [The five skills](/cli-reference/skills).
25
+ </Step>
26
+
19
27
  <Step title="Pick a dataset">
20
28
  The catalog lists every dataset you can run, with its versions.
21
29
 
@@ -28,7 +36,7 @@ This page takes you from an empty terminal to a downloaded trial: one API key, a
28
36
  </Step>
29
37
 
30
38
  <Step title="Start a job and watch it">
31
- Name a dataset, an agent, and a model. `--watch` streams the job's events until it finishes; `-l 5` caps the run at five tasks so the first job is small.
39
+ Name a dataset, an agent, and a model. `--watch` streams the job's events until it finishes; `-l 5` caps the run at five tasks and `--max-trial-spend 1` caps what each trial may spend on the model, so the first job is small.
32
40
 
33
41
  ```bash
34
42
  evolve run \
@@ -36,6 +44,7 @@ This page takes you from an empty terminal to a downloaded trial: one API key, a
36
44
  -a codex \
37
45
  -m gpt-6-astra \
38
46
  -l 5 \
47
+ --max-trial-spend 1 \
39
48
  --watch
40
49
  ```
41
50
 
@@ -67,19 +76,19 @@ This page takes you from an empty terminal to a downloaded trial: one API key, a
67
76
  </Step>
68
77
  </Steps>
69
78
 
70
- ## Next
79
+ ## Next, by what you want to do
71
80
 
72
81
  <Columns cols={2}>
73
- <Card title="Jobs" icon="list-checks" href="/core-concepts/jobs">
74
- Attempts, concurrency, retries, spend caps, and what to do with a finished job.
82
+ <Card title="Port a benchmark" icon="arrow-right-left" href="/core-concepts/tasks">
83
+ `evolve skills get create-adapter` turns it into Harbor-format tasks.
75
84
  </Card>
76
- <Card title="Analyze" icon="search" href="/core-concepts/analyze">
77
- Judge every trial's trace against a rubric with `--analyze`.
85
+ <Card title="Write a task" icon="file-plus" href="/core-concepts/tasks">
86
+ `evolve skills get create-task`, and `rewardkit` for its verifier.
78
87
  </Card>
79
- <Card title="Agents" icon="bot" href="/core-concepts/agents">
80
- The seven built-in harnesses, and how to bring your own.
88
+ <Card title="Publish a dataset" icon="upload" href="/core-concepts/datasets">
89
+ `evolve skills get publish`, then `evolve dataset publish`.
81
90
  </Card>
82
- <Card title="CLI reference" icon="terminal" href="/cli-reference/run">
83
- Every flag of `evolve run`.
91
+ <Card title="Upload a job you ran elsewhere" icon="folder-up" href="/core-concepts/upload">
92
+ `evolve upload <job dir>` brings a Harbor job into the dashboard.
84
93
  </Card>
85
94
  </Columns>
@@ -20,6 +20,18 @@ evolve run \
20
20
  --watch
21
21
  ```
22
22
 
23
+ New here? The [quick start](/getting-started/quick-start) takes five minutes.
24
+
25
+ ## The whole process
26
+
27
+ 1. [Get tasks](/core-concepts/tasks): a public dataset, your own, or a converted benchmark.
28
+ 2. [Check](/core-concepts/check) each task against a rubric before you run it.
29
+ 3. [Publish](/core-concepts/datasets) the tasks as a versioned dataset.
30
+ 4. [Run](/core-concepts/jobs) any agent and model on it; every trial streams live.
31
+ 5. [Read](/core-concepts/trial-outputs) every trial's tree and [analyze](/core-concepts/analyze) its trace.
32
+
33
+ A coding agent does all five from the same CLI with the [skills](/cli-reference/skills).
34
+
23
35
  <Columns cols={2}>
24
36
  <Card title="Quick start" icon="play" href="/getting-started/quick-start">
25
37
  Install the CLI, start a job, read its result.