@evolvingmachines/evolve 0.0.65 → 0.0.67-train-harness-wave-2026-09-25.20260926.e00a8a6d
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/dist/{chunk-DK2MGUPB.js → chunk-DI6LPW2B.js} +2 -2
- package/dist/cli/index.cjs +29 -29
- package/dist/cli/index.d.cts +3 -3
- package/dist/cli/index.d.ts +3 -3
- package/dist/cli/index.js +18 -18
- package/dist/index.cjs +64 -60
- package/dist/index.d.cts +297 -18
- package/dist/index.d.ts +297 -18
- package/dist/index.js +48 -44
- package/dist/{types-B_qmzlMk.d.cts → types-Dj0iMl5V.d.cts} +70 -90
- package/dist/{types-B_qmzlMk.d.ts → types-Dj0iMl5V.d.ts} +70 -90
- package/harness-capabilities.json +285 -0
- package/package.json +9 -6
- package/skills/create-adapter/SKILL.md +71 -26
- package/skills/create-task/SKILL.md +124 -66
- package/skills/evolve/SKILL.md +37 -20
- package/skills/evolve-evals/SKILL.md +142 -69
- package/skills/evolve-evals/references/cli-reference/agent.md +76 -0
- package/skills/evolve-evals/references/cli-reference/analysis.md +78 -0
- package/skills/evolve-evals/references/cli-reference/analyze.md +97 -0
- package/skills/evolve-evals/references/cli-reference/auth.md +96 -0
- package/skills/evolve-evals/references/cli-reference/check.md +157 -0
- package/skills/evolve-evals/references/cli-reference/dataset.md +154 -0
- package/skills/evolve-evals/references/cli-reference/filesystem.md +168 -0
- package/skills/evolve-evals/references/cli-reference/index.md +149 -0
- package/skills/evolve-evals/references/cli-reference/job.md +198 -0
- package/skills/evolve-evals/references/cli-reference/run.md +188 -0
- package/skills/evolve-evals/references/cli-reference/secrets.md +77 -0
- package/skills/evolve-evals/references/cli-reference/session.md +56 -0
- package/skills/evolve-evals/references/cli-reference/skill.md +52 -0
- package/skills/evolve-evals/references/cli-reference/skills.md +85 -0
- package/skills/evolve-evals/references/cli-reference/trial.md +115 -0
- package/skills/evolve-evals/references/cli-reference/upload.md +72 -0
- package/skills/evolve-evals/references/core-concepts/agents.md +101 -0
- package/skills/evolve-evals/references/core-concepts/analyze.md +143 -0
- package/skills/evolve-evals/references/core-concepts/check.md +116 -0
- package/skills/evolve-evals/references/core-concepts/compatibility.md +39 -0
- package/skills/evolve-evals/references/core-concepts/compose.md +103 -0
- package/skills/evolve-evals/references/core-concepts/datasets.md +234 -0
- package/skills/evolve-evals/references/core-concepts/index.md +52 -0
- package/skills/evolve-evals/references/core-concepts/jobs.md +223 -0
- package/skills/evolve-evals/references/core-concepts/models.md +73 -0
- package/skills/evolve-evals/references/core-concepts/multi-step.md +93 -0
- package/skills/evolve-evals/references/core-concepts/rewardkit.md +341 -0
- package/skills/evolve-evals/references/core-concepts/sandboxes.md +73 -0
- package/skills/evolve-evals/references/core-concepts/secrets.md +86 -0
- package/skills/evolve-evals/references/core-concepts/sharing.md +81 -0
- package/skills/evolve-evals/references/core-concepts/skills.md +100 -0
- package/skills/evolve-evals/references/core-concepts/task-config.md +169 -0
- package/skills/evolve-evals/references/core-concepts/task-environment.md +127 -0
- package/skills/evolve-evals/references/core-concepts/task-reference.md +191 -0
- package/skills/evolve-evals/references/core-concepts/task-verifiers.md +209 -0
- package/skills/evolve-evals/references/core-concepts/tasks.md +90 -0
- package/skills/evolve-evals/references/core-concepts/teams.md +128 -0
- package/skills/evolve-evals/references/core-concepts/trial-outputs.md +225 -0
- package/skills/evolve-evals/references/core-concepts/trials.md +129 -0
- package/skills/evolve-evals/references/core-concepts/upload-sdk-session.md +240 -0
- package/skills/evolve-evals/references/core-concepts/upload.md +166 -0
- package/skills/evolve-evals/references/dashboard/checks.md +54 -0
- package/skills/evolve-evals/references/dashboard/datasets.md +65 -0
- package/skills/evolve-evals/references/dashboard/jobs.md +106 -0
- package/skills/evolve-evals/references/dashboard/traces.md +54 -0
- package/skills/evolve-evals/references/dashboard/trial-viewer.md +77 -0
- package/skills/evolve-evals/references/getting-started/first-task.md +228 -0
- package/skills/evolve-evals/references/getting-started/installation.md +79 -0
- package/skills/evolve-evals/references/getting-started/quick-start.md +100 -0
- package/skills/evolve-evals/references/index.md +58 -0
- package/skills/evolve-evals/references/sdk/http.md +110 -0
- package/skills/evolve-evals/references/sdk/python.md +108 -0
- package/skills/evolve-evals/references/sdk/typescript.md +104 -0
- package/skills/evolve-evals/references/sdk-reference/agents.md +75 -0
- package/skills/evolve-evals/references/sdk-reference/analyses.md +125 -0
- package/skills/evolve-evals/references/sdk-reference/auth.md +171 -0
- package/skills/evolve-evals/references/sdk-reference/checks.md +146 -0
- package/skills/evolve-evals/references/sdk-reference/datasets.md +170 -0
- package/skills/evolve-evals/references/sdk-reference/error-codes.md +231 -0
- package/skills/evolve-evals/references/sdk-reference/errors.md +98 -0
- package/skills/evolve-evals/references/sdk-reference/filesystem.md +166 -0
- package/skills/evolve-evals/references/sdk-reference/index.md +161 -0
- package/skills/evolve-evals/references/sdk-reference/jobs.md +250 -0
- package/skills/evolve-evals/references/sdk-reference/meta.md +54 -0
- package/skills/evolve-evals/references/sdk-reference/methods/agents.md +154 -0
- package/skills/evolve-evals/references/sdk-reference/methods/analyses.md +211 -0
- package/skills/evolve-evals/references/sdk-reference/methods/auth.md +257 -0
- package/skills/evolve-evals/references/sdk-reference/methods/checks.md +462 -0
- package/skills/evolve-evals/references/sdk-reference/methods/datasets.md +687 -0
- package/skills/evolve-evals/references/sdk-reference/methods/filesystem.md +527 -0
- package/skills/evolve-evals/references/sdk-reference/methods/jobs.md +1023 -0
- package/skills/evolve-evals/references/sdk-reference/methods/meta.md +213 -0
- package/skills/evolve-evals/references/sdk-reference/methods/secrets.md +157 -0
- package/skills/evolve-evals/references/sdk-reference/methods/skills.md +125 -0
- package/skills/evolve-evals/references/sdk-reference/methods/trials.md +371 -0
- package/skills/evolve-evals/references/sdk-reference/secrets.md +104 -0
- package/skills/evolve-evals/references/sdk-reference/skills.md +66 -0
- package/skills/evolve-evals/references/sdk-reference/trials.md +164 -0
- package/skills/evolve-evals/references/sdk-reference/types.md +1255 -0
- package/skills/publish/SKILL.md +103 -79
- package/skills/rewardkit/SKILL.md +78 -23
- package/spec/openapi.yaml +164 -160
- package/skills/evolve-evals/references/cli-reference/agent.mdx +0 -79
- package/skills/evolve-evals/references/cli-reference/analysis.mdx +0 -160
- package/skills/evolve-evals/references/cli-reference/analyze.mdx +0 -84
- package/skills/evolve-evals/references/cli-reference/auth.mdx +0 -133
- package/skills/evolve-evals/references/cli-reference/check.mdx +0 -263
- package/skills/evolve-evals/references/cli-reference/dataset.mdx +0 -151
- package/skills/evolve-evals/references/cli-reference/index.mdx +0 -108
- package/skills/evolve-evals/references/cli-reference/job.mdx +0 -278
- package/skills/evolve-evals/references/cli-reference/run.mdx +0 -204
- package/skills/evolve-evals/references/cli-reference/secrets.mdx +0 -95
- package/skills/evolve-evals/references/cli-reference/skill.mdx +0 -61
- package/skills/evolve-evals/references/cli-reference/skills.mdx +0 -81
- package/skills/evolve-evals/references/cli-reference/trial.mdx +0 -181
- package/skills/evolve-evals/references/cli-reference/upload.mdx +0 -36
- package/skills/evolve-evals/references/core-concepts/agents.mdx +0 -106
- package/skills/evolve-evals/references/core-concepts/analyze.mdx +0 -86
- package/skills/evolve-evals/references/core-concepts/check.mdx +0 -64
- package/skills/evolve-evals/references/core-concepts/datasets.mdx +0 -138
- package/skills/evolve-evals/references/core-concepts/index.mdx +0 -42
- package/skills/evolve-evals/references/core-concepts/jobs.mdx +0 -151
- package/skills/evolve-evals/references/core-concepts/models.mdx +0 -68
- package/skills/evolve-evals/references/core-concepts/sandboxes.mdx +0 -52
- package/skills/evolve-evals/references/core-concepts/secrets.mdx +0 -41
- package/skills/evolve-evals/references/core-concepts/sharing.mdx +0 -35
- package/skills/evolve-evals/references/core-concepts/skills.mdx +0 -42
- package/skills/evolve-evals/references/core-concepts/tasks.mdx +0 -216
- package/skills/evolve-evals/references/core-concepts/teams.mdx +0 -36
- package/skills/evolve-evals/references/core-concepts/trial-outputs.mdx +0 -117
- package/skills/evolve-evals/references/core-concepts/trials.mdx +0 -103
- package/skills/evolve-evals/references/core-concepts/upload.mdx +0 -93
- package/skills/evolve-evals/references/dashboard/checks.mdx +0 -21
- package/skills/evolve-evals/references/dashboard/datasets.mdx +0 -24
- package/skills/evolve-evals/references/dashboard/jobs.mdx +0 -65
- package/skills/evolve-evals/references/dashboard/traces.mdx +0 -17
- package/skills/evolve-evals/references/dashboard/trial-viewer.mdx +0 -30
- package/skills/evolve-evals/references/getting-started/installation.mdx +0 -56
- package/skills/evolve-evals/references/getting-started/quick-start.mdx +0 -94
- package/skills/evolve-evals/references/index.mdx +0 -48
- package/skills/evolve-evals/references/sdk/python.mdx +0 -161
- package/skills/evolve-evals/references/sdk/typescript.mdx +0 -163
- package/skills/evolve-evals/references/sdk-reference/agents.mdx +0 -108
- package/skills/evolve-evals/references/sdk-reference/analyses.mdx +0 -222
- package/skills/evolve-evals/references/sdk-reference/auth.mdx +0 -116
- package/skills/evolve-evals/references/sdk-reference/checks.mdx +0 -300
- package/skills/evolve-evals/references/sdk-reference/datasets.mdx +0 -346
- package/skills/evolve-evals/references/sdk-reference/errors.mdx +0 -210
- package/skills/evolve-evals/references/sdk-reference/index.mdx +0 -60
- package/skills/evolve-evals/references/sdk-reference/jobs.mdx +0 -822
- package/skills/evolve-evals/references/sdk-reference/meta.mdx +0 -52
- package/skills/evolve-evals/references/sdk-reference/skills.mdx +0 -69
- package/skills/evolve-evals/references/sdk-reference/trials.mdx +0 -302
- package/skills/evolve-evals/references/sdk-reference/types.mdx +0 -274
- package/skills/evolve-evals/references/snippets/global-options.mdx +0 -13
- package/skills/evolve-evals/references/snippets/list-flags.mdx +0 -23
|
@@ -1,263 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
title: "evolve check"
|
|
3
|
-
description: "Check task quality against a rubric, and read checks back."
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
import GlobalOptions from '/snippets/global-options.mdx';
|
|
7
|
-
import ListFlags from '/snippets/list-flags.mdx';
|
|
8
|
-
|
|
9
|
-
```bash
|
|
10
|
-
evolve check [<path>] [options]
|
|
11
|
-
```
|
|
12
|
-
|
|
13
|
-
Check task quality against a rubric, server-side. `<path>` is one task directory or a directory of task directories, uploaded as one archive (`upload_too_large` past the published ceiling); `-d` names a published dataset instead, one or the other. Without `--watch` the command prints the accepted check with the hint `Follow it with: evolve check show <id>`; `--json` prints the accepted check body. Exit 0 on an accepted check; with `--watch`, and on `check show`, exit 1 when any task errored.
|
|
14
|
-
|
|
15
|
-
```bash
|
|
16
|
-
evolve check ./tasks \
|
|
17
|
-
-i 'abs-*' \
|
|
18
|
-
-l 5 \
|
|
19
|
-
--watch
|
|
20
|
-
evolve check --show-defaults
|
|
21
|
-
```
|
|
22
|
-
|
|
23
|
-
## Options
|
|
24
|
-
|
|
25
|
-
<ParamField path="-d, --dataset" type="name[@version]">
|
|
26
|
-
Check a published dataset's tasks instead of a local path. The checker reads the version's retained task package, and the job page's CHECK tab lists these checks by task.
|
|
27
|
-
</ParamField>
|
|
28
|
-
|
|
29
|
-
<ParamField path="--name" type="name">
|
|
30
|
-
A name for the check, Harbor's `--job-name`. Omitted, the accept timestamp in Harbor's shape, `YYYY-MM-DD__HH-MM-SS`.
|
|
31
|
-
</ParamField>
|
|
32
|
-
|
|
33
|
-
<ParamField path="-m, --model" type="name" default="openrouter/deepseek/deepseek-v4.1-flash">
|
|
34
|
-
Model the checker runs. `fireworks/deepseek-v4.1-flash` is the same model on its Fireworks route. Must be on the `claude` roster.
|
|
35
|
-
</ParamField>
|
|
36
|
-
|
|
37
|
-
<ParamField path="--effort" type="value">
|
|
38
|
-
Reasoning effort the checker runs at. Defaults to the per-model default.
|
|
39
|
-
</ParamField>
|
|
40
|
-
|
|
41
|
-
<ParamField path="-r, --rubric" type="path">
|
|
42
|
-
Rubric file: TOML, YAML or JSON in the `{criteria: [{name, description, guidance}]}` shape. Defaults to the platform's check rubric, eleven criteria.
|
|
43
|
-
</ParamField>
|
|
44
|
-
|
|
45
|
-
<ParamField path="-p, --prompt" type="path">
|
|
46
|
-
Prompt file for the evaluator agent. Its text replaces the built-in check prompt, with `{task_path}`, `{file_tree}` and `{criteria_guidance}` rendered. Defaults to the built-in prompt.
|
|
47
|
-
</ParamField>
|
|
48
|
-
|
|
49
|
-
<ParamField path="-e, --env" type="provider">
|
|
50
|
-
Sandbox provider the checker runs on. Defaults to the platform's analysis default.
|
|
51
|
-
</ParamField>
|
|
52
|
-
|
|
53
|
-
<ParamField path="--show-defaults">
|
|
54
|
-
Print the defaults a check runs under when you name none: model, effort, provider, the rubric's criteria and the built-in prompt template, then exit. With `--json`, one object. Takes no path and no other check flag.
|
|
55
|
-
</ParamField>
|
|
56
|
-
|
|
57
|
-
<ParamField path="-n, --n-concurrent" type="n">
|
|
58
|
-
Max concurrent task checks, beneath the organization's ceiling. Defaults to the ceiling.
|
|
59
|
-
</ParamField>
|
|
60
|
-
|
|
61
|
-
<ParamField path="-i, --include-task-name" type="glob">
|
|
62
|
-
Only check tasks matching the glob. Repeatable.
|
|
63
|
-
</ParamField>
|
|
64
|
-
|
|
65
|
-
<ParamField path="-x, --exclude-task-name" type="glob">
|
|
66
|
-
Skip tasks matching the glob. Repeatable.
|
|
67
|
-
</ParamField>
|
|
68
|
-
|
|
69
|
-
<ParamField path="-l, --n-tasks" type="n">
|
|
70
|
-
Max tasks to check, after the globs, in sorted directory order.
|
|
71
|
-
</ParamField>
|
|
72
|
-
|
|
73
|
-
<ParamField path="--watch">
|
|
74
|
-
Poll until every task check settles, then print the report. One task prints its label line, a `CHECK | OUTCOME | EXPLANATION` table with the evidence beneath each verdict, and the cost; several tasks print a `TASK | LABEL | PASS | FAIL | N/A | UNKNOWN | COST ($) | TASK CHECK ID` table with each failed task's reason below it. With `--json`, NDJSON envelopes: `check.accepted`, `check.progress` per change, `check.final`.
|
|
75
|
-
</ParamField>
|
|
76
|
-
|
|
77
|
-
<ParamField path="-q, --quiet">
|
|
78
|
-
With `--watch`: suppress the progress lines and print the final report only.
|
|
79
|
-
</ParamField>
|
|
80
|
-
|
|
81
|
-
<GlobalOptions />
|
|
82
|
-
|
|
83
|
-
## Reading checks back
|
|
84
|
-
|
|
85
|
-
The same word, followed by `list`, `show`, `trace`, `download`, `share`, `unshare` or `shares`, reads checks back or shares them. A task directory literally named like one of these verbs is written `./list`.
|
|
86
|
-
|
|
87
|
-
### check list
|
|
88
|
-
|
|
89
|
-
```bash
|
|
90
|
-
evolve check list [options]
|
|
91
|
-
```
|
|
92
|
-
|
|
93
|
-
List your task quality checks, newest first.
|
|
94
|
-
|
|
95
|
-
<ListFlags />
|
|
96
|
-
|
|
97
|
-
<ParamField path="--scope" type="my | shared">
|
|
98
|
-
Visibility scope: `my`, what you created (the default), or `shared`, your organizations' rows that teammates created plus the checks shared with your address.
|
|
99
|
-
</ParamField>
|
|
100
|
-
|
|
101
|
-
<ParamField path="--status" type="s1,s2,...">
|
|
102
|
-
Filter by check status: `queued`, `running`, `completed`.
|
|
103
|
-
</ParamField>
|
|
104
|
-
|
|
105
|
-
<ParamField path="-d, --dataset" type="name[@version]">
|
|
106
|
-
Only checks on this dataset. A bare name is every version, `name@version` one version. A value with an empty half (`name@`, `@version`) is refused as `invalid_input` on `dataset`.
|
|
107
|
-
</ParamField>
|
|
108
|
-
|
|
109
|
-
```bash
|
|
110
|
-
evolve check list --status running
|
|
111
|
-
evolve check list --dataset terminal-bench-4@4.0
|
|
112
|
-
```
|
|
113
|
-
|
|
114
|
-
### check show
|
|
115
|
-
|
|
116
|
-
```bash
|
|
117
|
-
evolve check show <check-id>
|
|
118
|
-
```
|
|
119
|
-
|
|
120
|
-
Show one task quality check in full, with one result per task. The `visibility` row is `PRIVATE` or `LINK` — whether the check's share link is on ([check shares](#check-shares) prints the link and the addresses).
|
|
121
|
-
|
|
122
|
-
### check trace
|
|
123
|
-
|
|
124
|
-
```bash
|
|
125
|
-
evolve check trace <task-check-id> [options]
|
|
126
|
-
```
|
|
127
|
-
|
|
128
|
-
Print the checker's own parsed transcript for one task check.
|
|
129
|
-
|
|
130
|
-
<ParamField path="--since" type="n">
|
|
131
|
-
Skip the first N events. To resume, pass the count you already hold.
|
|
132
|
-
</ParamField>
|
|
133
|
-
|
|
134
|
-
### check download
|
|
135
|
-
|
|
136
|
-
```bash
|
|
137
|
-
evolve check download <check-id | task-check-id> [options]
|
|
138
|
-
```
|
|
139
|
-
|
|
140
|
-
Save a check as a check folder (`check_report.json` plus one folder per task) by its check id, one task check's folder by its task check id, or stream one task check's artifact.
|
|
141
|
-
|
|
142
|
-
<ParamField path="-o, --output-dir" type="dir" default="checks/">
|
|
143
|
-
Directory to save under. A check id lands in `<dir>/check-<id>/`, a task check id in `<dir>/check-<task>__<7 chars>/`, the run's own folder name.
|
|
144
|
-
</ParamField>
|
|
145
|
-
|
|
146
|
-
<ParamField path="--overwrite">
|
|
147
|
-
Replace an existing folder of the same name.
|
|
148
|
-
</ParamField>
|
|
149
|
-
|
|
150
|
-
<ParamField path="--stream" type="artifact">
|
|
151
|
-
Print one artifact of a task check to stdout instead of saving: `task-check` (the result document), `trace-parsed`, `trace-stdout`, `trace-stderr` or `agent-home`.
|
|
152
|
-
</ParamField>
|
|
153
|
-
|
|
154
|
-
<ParamField path="--since" type="n">
|
|
155
|
-
With `--stream trace-parsed`: skip the first N events.
|
|
156
|
-
</ParamField>
|
|
157
|
-
|
|
158
|
-
```bash
|
|
159
|
-
evolve check download 3f9a1c2e-… -o checks/
|
|
160
|
-
```
|
|
161
|
-
|
|
162
|
-
### check share
|
|
163
|
-
|
|
164
|
-
```bash
|
|
165
|
-
evolve check share <check-id> --link
|
|
166
|
-
evolve check share <check-id> --email <address> [--email <address>...]
|
|
167
|
-
```
|
|
168
|
-
|
|
169
|
-
Share a check you created, the way [job share](/cli-reference/job#job-share) shares a job. `--link` turns on the check's unlisted link and prints it; running it again prints the same link. Whoever opens the link reads the check's report and downloads its check folder, without signing in. `--email` shares the check with an address: that person gets an email with a link and reads the check after signing in with the same address (a person without an account gets a sign-up link in the same email; an address already on the waitlist or holding an invite gets the share email without a sign-up link and uses that invite instead). Both flags may ride one command. Someone the check is shared with reads it — the report, each task check's files and logs, the download — and finds it under `evolve check list --scope shared`; they only read, and can never stop one of its task checks. Only the check's creator may share it; another member of the organization is refused with `org_forbidden`.
|
|
170
|
-
|
|
171
|
-
```bash
|
|
172
|
-
evolve check share 3f9a1c2e-… --link
|
|
173
|
-
evolve check share 3f9a1c2e-… --email alice@example.org
|
|
174
|
-
```
|
|
175
|
-
|
|
176
|
-
### check unshare
|
|
177
|
-
|
|
178
|
-
```bash
|
|
179
|
-
evolve check unshare <check-id> --link
|
|
180
|
-
evolve check unshare <check-id> --email <address>
|
|
181
|
-
```
|
|
182
|
-
|
|
183
|
-
Take a share back. `--link` turns the link off at once — the old link stops working, and a later `check share --link` prints a new one. `--email` removes that address's share. Both are safe to repeat.
|
|
184
|
-
|
|
185
|
-
### check shares
|
|
186
|
-
|
|
187
|
-
```bash
|
|
188
|
-
evolve check shares <check-id>
|
|
189
|
-
```
|
|
190
|
-
|
|
191
|
-
Print who the check is shared with: its visibility, the link while it is on, and every address with who shared it and when. `--json` prints the same as `{"visibility", "link": {"enabled", "url"}, "emails": [{"email", "shared_by", "created_at"}]}`.
|
|
192
|
-
|
|
193
|
-
## check files
|
|
194
|
-
|
|
195
|
-
A task check's run, by its task check id (`results[].id` on `evolve check show`).
|
|
196
|
-
|
|
197
|
-
|
|
198
|
-
```bash
|
|
199
|
-
evolve check files <status|ls|cat|search|changes|archive> <task-check-id> [path] [options]
|
|
200
|
-
```
|
|
201
|
-
|
|
202
|
-
The run's file system: while the box runs, `files` reads it as it changes; after the run, the same commands read the kept tree. `files status` says which you get.
|
|
203
|
-
|
|
204
|
-
- `files status <task-check-id>` — `live`, `captured`, `capturing` (the kept tree is still being written) or `none`; the box, how change events arrive (`watcher`), and the capture record (`ready`, `incomplete` with what was left out, or `failed`).
|
|
205
|
-
- `files ls <task-check-id> [path]` — one folder, sorted by name (default `/`). Each row shows the type, mode, size, mtime and name; `created (agent)` or `modified (verifier)` marks what the run changed and when, `image only` marks a file the run never touched (it lists but does not open on the kept tree).
|
|
206
|
-
- `files cat <task-check-id> <path>` — the raw bytes of one file, to stdout. `--range bytes=a-b` (or `bytes=a-`, `bytes=-n`) reads a slice.
|
|
207
|
-
- `files search <task-check-id> <text>` — every line matching the text, as `path:line: snippet`. `--path` narrows the search to one folder (the whole box takes seconds); `--regex` treats the text as a pattern; a `(truncated)` note means there was more.
|
|
208
|
-
- `files changes <task-check-id>` — the files the run created, modified or removed, with the phase (`--phase agent|verifier`).
|
|
209
|
-
- `files archive <task-check-id>` — a `.tar.gz` of one subtree (`--path`, default the whole tree) to stdout, or saved under `-o <dir>`. The kept tree's archive lists the files it does not carry in `MANIFEST.tsv`; while the box runs, a subtree too large to read out in time is refused (`feature_unsupported`).
|
|
210
|
-
|
|
211
|
-
<ParamField path="--source" type="live|capture">
|
|
212
|
-
Force the source. Without it, whichever the run has. A source that is not there is refused with `filesystem_state`.
|
|
213
|
-
</ParamField>
|
|
214
|
-
|
|
215
|
-
<ParamField path="-l, --limit" type="n">
|
|
216
|
-
Page size on `ls`, `search` and `changes` (`ls`, `changes` default 500; `search` default 200; max 1000).
|
|
217
|
-
</ParamField>
|
|
218
|
-
|
|
219
|
-
<ParamField path="--cursor" type="c">
|
|
220
|
-
Resume a paged `ls` or `changes` after the previous page's `next_cursor`.
|
|
221
|
-
</ParamField>
|
|
222
|
-
|
|
223
|
-
```bash
|
|
224
|
-
evolve check files ls 3f9a1c2e-… /app/work
|
|
225
|
-
evolve check files search 3f9a1c2e-… 'permission denied' --path /app/work
|
|
226
|
-
evolve check files archive 3f9a1c2e-… --path /app/work -o archives/
|
|
227
|
-
```
|
|
228
|
-
|
|
229
|
-
## check logs
|
|
230
|
-
|
|
231
|
-
```bash
|
|
232
|
-
evolve check logs <task-check-id> --stream <name> [--follow] [options]
|
|
233
|
-
```
|
|
234
|
-
|
|
235
|
-
The box's own streams: `agent` (the harness's stdout and stderr), `verifier` and `system` (the box's system log, only when the job asked for it with `--system-log`); `setup` and `metrics` are named but not recorded today, so their pages are empty and say why. One page by default; `--follow` keeps printing lines as they arrive while the box lives, and returns once it is gone.
|
|
236
|
-
|
|
237
|
-
<ParamField path="--stream" type="name" required>
|
|
238
|
-
`agent`, `verifier`, `system`, `setup` or `metrics`.
|
|
239
|
-
</ParamField>
|
|
240
|
-
|
|
241
|
-
<ParamField path="-f, --follow">
|
|
242
|
-
Keep printing new lines while the box lives.
|
|
243
|
-
</ParamField>
|
|
244
|
-
|
|
245
|
-
<ParamField path="-l, --limit" type="n" default="1000">
|
|
246
|
-
Lines per page, max 1000.
|
|
247
|
-
</ParamField>
|
|
248
|
-
|
|
249
|
-
<ParamField path="--cursor" type="seq">
|
|
250
|
-
Resume after this line.
|
|
251
|
-
</ParamField>
|
|
252
|
-
|
|
253
|
-
```bash
|
|
254
|
-
evolve check logs 3f9a1c2e-… --stream agent --follow
|
|
255
|
-
```
|
|
256
|
-
|
|
257
|
-
## check procs
|
|
258
|
-
|
|
259
|
-
```bash
|
|
260
|
-
evolve check procs <task-check-id>
|
|
261
|
-
```
|
|
262
|
-
|
|
263
|
-
The processes running in the box right now. Live only: once the box is gone the command is refused with `filesystem_state`.
|
|
@@ -1,151 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
title: "evolve dataset"
|
|
3
|
-
description: "Browse and publish the dataset catalog."
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
import GlobalOptions from '/snippets/global-options.mdx';
|
|
7
|
-
import ListFlags from '/snippets/list-flags.mdx';
|
|
8
|
-
|
|
9
|
-
```bash
|
|
10
|
-
evolve dataset <command> [options]
|
|
11
|
-
```
|
|
12
|
-
|
|
13
|
-
## dataset list
|
|
14
|
-
|
|
15
|
-
```bash
|
|
16
|
-
evolve dataset list [options]
|
|
17
|
-
```
|
|
18
|
-
|
|
19
|
-
List the dataset catalog.
|
|
20
|
-
|
|
21
|
-
<ListFlags />
|
|
22
|
-
|
|
23
|
-
<ParamField path="--search" type="text">
|
|
24
|
-
Free-text filter over name and description.
|
|
25
|
-
</ParamField>
|
|
26
|
-
|
|
27
|
-
## dataset show
|
|
28
|
-
|
|
29
|
-
```bash
|
|
30
|
-
evolve dataset show <name[@version]> [options]
|
|
31
|
-
```
|
|
32
|
-
|
|
33
|
-
Show one dataset: its versions, its manifest identity, its tasks with a NOTES column, the providers each task can run on, and the failed tasks of the shown version with their reasons. One line reports a git dataset whose upstream ref has moved; it never appears in `--json`, where the same fact is the `upstream` field.
|
|
34
|
-
|
|
35
|
-
<ParamField path="-l, --limit" type="n">
|
|
36
|
-
Task-list page size.
|
|
37
|
-
</ParamField>
|
|
38
|
-
|
|
39
|
-
<ParamField path="--cursor" type="c">
|
|
40
|
-
Resume the task list from a cursor.
|
|
41
|
-
</ParamField>
|
|
42
|
-
|
|
43
|
-
```bash
|
|
44
|
-
evolve dataset show terminal-bench-4@4.0
|
|
45
|
-
```
|
|
46
|
-
|
|
47
|
-
## dataset check
|
|
48
|
-
|
|
49
|
-
```bash
|
|
50
|
-
evolve dataset check <dir>
|
|
51
|
-
```
|
|
52
|
-
|
|
53
|
-
Pre-flight a local corpus. A dry run: nothing is uploaded, nothing is written. Prints a verdict per task and exits 1 on any refusal.
|
|
54
|
-
|
|
55
|
-
## dataset publish
|
|
56
|
-
|
|
57
|
-
```bash
|
|
58
|
-
evolve dataset publish [options]
|
|
59
|
-
```
|
|
60
|
-
|
|
61
|
-
Publish a dataset version from a git source, a local directory, or a fetchable source: a public tarball URL, or a public hub package. A name owned by someone else is refused before the bytes move, and an archive over the published ceiling is refused with `import_too_large`.
|
|
62
|
-
|
|
63
|
-
A directory upload prints `upload <sent>/<total> (P%)` once per 10% of the archive; with `--json --watch` the same steps are `upload.progress` events (`sent_bytes`, `total_bytes`, `elapsed_sec`). A large upload prints `Registered import <id>` early and reads `QUEUED (receiving)` until the corpus has arrived. Once the build settles, `--watch` prints every task's outcome and ends with `built N of M tasks`.
|
|
64
|
-
|
|
65
|
-
<ParamField path="--git" type="url">
|
|
66
|
-
Git repository URL. With `--ref`.
|
|
67
|
-
</ParamField>
|
|
68
|
-
|
|
69
|
-
<ParamField path="--ref" type="ref">
|
|
70
|
-
Pinned git ref: a full 40-hex commit sha, or a tag, resolved to its commit at publish and verified at import. Branch names are refused as `unpinned_git_ref`. With `--git`.
|
|
71
|
-
</ParamField>
|
|
72
|
-
|
|
73
|
-
<ParamField path="--path" type="subfolder">
|
|
74
|
-
Repository subfolder holding the corpus. With `--git`; only that folder is imported.
|
|
75
|
-
</ParamField>
|
|
76
|
-
|
|
77
|
-
<ParamField path="--dir" type="path">
|
|
78
|
-
Local corpus directory, tarred and uploaded. Pre-flighted first; see `--skip-preflight`.
|
|
79
|
-
</ParamField>
|
|
80
|
-
|
|
81
|
-
<ParamField path="--from" type="url | hub:org/name[@ref]">
|
|
82
|
-
Fetchable source the server pulls itself, with no local bytes: a public https tarball URL, or `hub:org/name[@ref]`, a public Harbor hub package. The ref is the latest tag by default, a revision number, or `sha256:<digest>`; it is resolved and digest-pinned when the publish is accepted.
|
|
83
|
-
</ParamField>
|
|
84
|
-
|
|
85
|
-
<ParamField path="--name" type="dataset">
|
|
86
|
-
Catalog dataset name to create or extend. Optional with `--dir` when the corpus carries a `dataset.toml` manifest, and with `--from hub:…`, which defaults to the package's short name. Required with `--git` and `--from <url>`.
|
|
87
|
-
</ParamField>
|
|
88
|
-
|
|
89
|
-
<ParamField path="--version" type="v">
|
|
90
|
-
Version label for the published version. Optional with `--dir` when `dataset.toml` declares one, and with `--from hub:…`, which defaults to the resolved revision. Required with `--git` and `--from <url>`.
|
|
91
|
-
</ParamField>
|
|
92
|
-
|
|
93
|
-
<ParamField path="--org" type="name">
|
|
94
|
-
Organization a new dataset lands in, so your teammates can run it. Omitted, the default set with `evolve auth org use`, else your personal organization; `--org personal` picks your personal organization for this one publish over that default. An existing dataset stays where it is; naming a different organization for it is refused.
|
|
95
|
-
</ParamField>
|
|
96
|
-
|
|
97
|
-
<ParamField path="--watch">
|
|
98
|
-
Poll until the publish settles: the version `READY` (built and active) or `FAILED`.
|
|
99
|
-
</ParamField>
|
|
100
|
-
|
|
101
|
-
<ParamField path="--skip-preflight">
|
|
102
|
-
Upload without the pre-flight check. A refused task then lands `FAILED` at import instead of being caught here.
|
|
103
|
-
</ParamField>
|
|
104
|
-
|
|
105
|
-
```bash
|
|
106
|
-
evolve dataset publish --from hub:cookbook/hello-world --watch
|
|
107
|
-
evolve dataset publish --dir ./corpus --name my-swe --version 1.0 --org acme
|
|
108
|
-
```
|
|
109
|
-
|
|
110
|
-
## dataset watch
|
|
111
|
-
|
|
112
|
-
```bash
|
|
113
|
-
evolve dataset watch <name | import-id>
|
|
114
|
-
```
|
|
115
|
-
|
|
116
|
-
Re-attach to a publish and follow it to `READY` or `FAILED`, the same follow `dataset publish --watch` renders. Works after the CLI exited, or from another machine. A name with no live publish is refused, naming the newest settled import. With `--json` the same events stream, opened with `import.attached`; the transfer's own `upload.progress` lines belong to the publishing process and never appear here.
|
|
117
|
-
|
|
118
|
-
## dataset download
|
|
119
|
-
|
|
120
|
-
```bash
|
|
121
|
-
evolve dataset download <name[@version]> [options]
|
|
122
|
-
```
|
|
123
|
-
|
|
124
|
-
Download the original corpus package. Owner only.
|
|
125
|
-
|
|
126
|
-
<ParamField path="-o, --output-dir" type="dir">
|
|
127
|
-
Directory to save into. Defaults to the current directory.
|
|
128
|
-
</ParamField>
|
|
129
|
-
|
|
130
|
-
## dataset files
|
|
131
|
-
|
|
132
|
-
```bash
|
|
133
|
-
evolve dataset files <status|ls|cat> <name@version> <task> [path] [options]
|
|
134
|
-
```
|
|
135
|
-
|
|
136
|
-
A published task's files, exactly as the platform stored them and as every trial ran against them: `files ls` lists one folder of the task directory (default `/`: `instruction.md`, `task.toml`, `environment/`, `tests/`), `files cat` prints one file's bytes (`--range bytes=a-b` reads a slice), `files status` says whether the version keeps its package (`package_retained`). A version imported before packages were kept is refused with `task_package_not_retained`. The version must be named: `name@version`.
|
|
137
|
-
|
|
138
|
-
```bash
|
|
139
|
-
evolve dataset files ls my-swe@1.0 abs-1 /tests
|
|
140
|
-
evolve dataset files cat my-swe@1.0 abs-1 /tests/test.sh
|
|
141
|
-
```
|
|
142
|
-
|
|
143
|
-
## dataset activate
|
|
144
|
-
|
|
145
|
-
```bash
|
|
146
|
-
evolve dataset activate <name> <version>
|
|
147
|
-
```
|
|
148
|
-
|
|
149
|
-
Make a `READY` version the dataset's active version.
|
|
150
|
-
|
|
151
|
-
<GlobalOptions />
|
|
@@ -1,108 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
title: "CLI reference"
|
|
3
|
-
description: "Every verb of the evolve command, from its own help output."
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
```bash
|
|
7
|
-
evolve <command> [options]
|
|
8
|
-
```
|
|
9
|
-
|
|
10
|
-
The `evolve` command starts jobs and reads everything back. Every verb accepts `--json` for machine-readable output, and `-h` or `--help` prints its flags under short headings, with defaults, and one to three examples.
|
|
11
|
-
|
|
12
|
-
## Verbs
|
|
13
|
-
|
|
14
|
-
| Verb | What it does |
|
|
15
|
-
| --- | --- |
|
|
16
|
-
| [`run`](/cli-reference/run) | Start a job. Add `--watch` to follow it. The short form of `job start`. |
|
|
17
|
-
| [`analyze`](/cli-reference/analyze) | Analyze a finished job's trial traces against a rubric. |
|
|
18
|
-
| [`check`](/cli-reference/check) | Check task quality against a rubric. |
|
|
19
|
-
| [`upload`](/cli-reference/upload) | Upload a job directory as a finished job. |
|
|
20
|
-
|
|
21
|
-
## Command groups
|
|
22
|
-
|
|
23
|
-
| Group | Commands |
|
|
24
|
-
| --- | --- |
|
|
25
|
-
| [`skills`](/cli-reference/skills) | `list`, `get`, `path`, `install` |
|
|
26
|
-
| [`job`](/cli-reference/job) | `start`, `list`, `show`, `trials`, `tasks`, `compare`, `cancel`, `delete`, `stop`, `resume`, `retry`, `regrade`, `imports`, `import`, `download`, `grep` |
|
|
27
|
-
| [`trial`](/cli-reference/trial) | `show`, `trace`, `download`, `retry`, `regrade`, `stop` |
|
|
28
|
-
| [`analysis`](/cli-reference/analysis) | `list`, `show`, `trace`, `download` |
|
|
29
|
-
| [`check`](/cli-reference/check) | `list`, `show`, `trace`, `download` |
|
|
30
|
-
| [`dataset`](/cli-reference/dataset) | `list`, `show`, `check`, `publish`, `watch`, `download`, `activate` |
|
|
31
|
-
| [`skill`](/cli-reference/skill) | `list`, `upload`, `show`, `delete` |
|
|
32
|
-
| [`agent`](/cli-reference/agent) | `list`, `show`, `add`, `remove` |
|
|
33
|
-
| [`auth`](/cli-reference/auth) | `status`, `org list`, `org show` |
|
|
34
|
-
| [`secrets`](/cli-reference/secrets) | `set`, `list`, `delete` |
|
|
35
|
-
|
|
36
|
-
The singular noun is canonical; `jobs`, `trials`, `analyses` and `datasets` are hidden aliases, as are `secret` for `secrets` and `ls` for `list`. `skills` is not the plural of `skill`: it is the group that serves the bundled skills to your coding agent. `agents` is not an alias: it is reserved for the managed-agents CLI and refuses, pointing at `evolve agent`.
|
|
37
|
-
|
|
38
|
-
## Global options
|
|
39
|
-
|
|
40
|
-
These work on every command.
|
|
41
|
-
|
|
42
|
-
<ParamField path="--json">
|
|
43
|
-
Machine-readable JSON output. A refusal then prints one `{"error": {...}}` object on stdout with the server's `code`, `message`, `param`, `details`, `retryAfterSec` and `request_id` when present.
|
|
44
|
-
</ParamField>
|
|
45
|
-
|
|
46
|
-
<ParamField path="--api-key" type="key">
|
|
47
|
-
API key. Defaults to `$EVOLVE_API_KEY`.
|
|
48
|
-
</ParamField>
|
|
49
|
-
|
|
50
|
-
<ParamField path="--base-url" type="url">
|
|
51
|
-
API base URL. Defaults to the Evolve dashboard API.
|
|
52
|
-
</ParamField>
|
|
53
|
-
|
|
54
|
-
<ParamField path="-v, --version">
|
|
55
|
-
Print the CLI version.
|
|
56
|
-
</ParamField>
|
|
57
|
-
|
|
58
|
-
## Ids
|
|
59
|
-
|
|
60
|
-
Every verb that takes an id (`job`, `trial`, `analysis`, `check` with a check id or a task check id, `skill`) also takes an unambiguous prefix of at least 8 characters. A prefix that matches nothing or more than one row is refused naming the candidates. A full id is never prefix-matched, `name:<skill-name>` is a name, and a job import id is taken as printed.
|
|
61
|
-
|
|
62
|
-
A trial prefix is resolved against every trial of every job in scope, so on an account with many jobs the lookup can hit the rate limit; the full id, which `evolve job trials <job>` prints, always works.
|
|
63
|
-
|
|
64
|
-
## Output
|
|
65
|
-
|
|
66
|
-
Human tables on a TTY, tab-separated rows when piped, `--json` for the machine shape (NDJSON under `--watch`), and `-q` for ids only. `--no-trunc` disables cell truncation and `--no-headers` drops the header row from piped output. A money cell carries its state: `$0.06` is final, `at least $0.06` is still being written, `-` means not measured yet.
|
|
67
|
-
|
|
68
|
-
`--columns` chooses and orders list columns, comma-separated; `--columns help` prints the keys. For `job list` they are `id`, `name`, `status`, `datasets`, `agents`, `trials`, `spent`, `started`; for `job trials` `task`, `agent`, `attempt`, `status`, `reward`, `spent`, `tokens`, `gpu`, `id`; for `analysis list` `id`, `status`, `task`, `job`, `trial`, `model`, `attempts`, `spent`, `created`, `finished`.
|
|
69
|
-
|
|
70
|
-
## Listing flags
|
|
71
|
-
|
|
72
|
-
Every `list` command, and `job trials`, `job tasks` and `job imports`, share one set of paging and formatting flags. Two take only the formatting flags, not `-l` or `--cursor`: `auth org list` and `secrets list`.
|
|
73
|
-
|
|
74
|
-
<ParamField path="-l, --limit" type="n">
|
|
75
|
-
Page size.
|
|
76
|
-
</ParamField>
|
|
77
|
-
|
|
78
|
-
<ParamField path="--cursor" type="c">
|
|
79
|
-
Resume from a page cursor.
|
|
80
|
-
</ParamField>
|
|
81
|
-
|
|
82
|
-
<ParamField path="--columns" type="keys | all | help">
|
|
83
|
-
Choose and order columns, comma-separated. `help` lists the keys.
|
|
84
|
-
</ParamField>
|
|
85
|
-
|
|
86
|
-
<ParamField path="-q, --quiet">
|
|
87
|
-
Print only ids, one per line, for piping.
|
|
88
|
-
</ParamField>
|
|
89
|
-
|
|
90
|
-
<ParamField path="--no-trunc">
|
|
91
|
-
Full cell content instead of one-line truncation.
|
|
92
|
-
</ParamField>
|
|
93
|
-
|
|
94
|
-
<ParamField path="--no-headers">
|
|
95
|
-
Omit the header row in piped (TSV) output.
|
|
96
|
-
</ParamField>
|
|
97
|
-
|
|
98
|
-
## Scope
|
|
99
|
-
|
|
100
|
-
`--scope` on `job list`, `analysis list`, `check list`, `session list`, `skill list` and `agent list` is `my`, what you created and the default; `shared`, your organizations' rows that teammates created; or `org`, everything in your organizations, your own rows included. `all` is refused. Every id printed under any scope resolves on its `show`, `trace` and `download`; an id outside your organizations answers the noun's not-found code, exactly like an id that does not exist. The one exception is a registered agent, addressed by name rather than id: `agent show <name>` opens your own registration of that name, else a teammate's in one of your organizations there — it is usable in a job's `agents[].name`, not on `show`.
|
|
101
|
-
|
|
102
|
-
## Values
|
|
103
|
-
|
|
104
|
-
A flag's value may begin with `-`, such as `-x '-*'` or a negative number. Only a token that spells another flag of the same command is refused, and the refusal shows the `--flag=value` form. A typo in `--stream`, `--status`, `--scope` or `run`'s `-e` is a usage error naming the legal values, before any request.
|
|
105
|
-
|
|
106
|
-
## Exit codes
|
|
107
|
-
|
|
108
|
-
`0` on success (with `--watch`: the job `COMPLETED`, a publish settled `READY`, every analysis completed, no task check errored). `1` on a runtime or API failure (with `--watch`: `FAILED` or `CANCELLED`, a failed analysis, an errored task check). `2` on a usage error, and on a quota refusal, which prints `Launch quota exceeded:` with the server's sentence. A rate limit prints one line with the wait and exits 1; the SDK's watch loops wait it out on their own.
|