@atolis-hq/wake 0.3.0 → 0.3.1
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.
- package/README.md +5 -5
- package/dist/src/bootstrap/initialise.js +13 -11
- package/dist/src/bootstrap/version.js +1 -1
- package/package.json +1 -1
- package/templates/SETUP.md +0 -115
package/README.md
CHANGED
|
@@ -108,8 +108,8 @@ For more detail, see [docs/vision.md](docs/vision.md) and
|
|
|
108
108
|
provider.
|
|
109
109
|
- **Human resumption.** A human can pick up the exact local agent session when a
|
|
110
110
|
direct terminal intervention is the best way forward.
|
|
111
|
-
- **Operator correlation escape hatch.** `wake correlate <
|
|
112
|
-
<
|
|
111
|
+
- **Operator correlation escape hatch.** `wake correlate <resourceUri>
|
|
112
|
+
<workItemId>` lets an operator hand-declare that a resource (a PR, a Slack
|
|
113
113
|
thread, etc.) belongs to an existing work item when nothing detected the
|
|
114
114
|
link automatically. See [docs/configuration.md](docs/configuration.md).
|
|
115
115
|
|
|
@@ -138,10 +138,10 @@ adapters currently exist for:
|
|
|
138
138
|
- **[Codex](https://openai.com/codex/)**
|
|
139
139
|
- **[Cursor](https://cursor.com/cli)**
|
|
140
140
|
|
|
141
|
-
Each runner sits behind the same
|
|
141
|
+
Each runner sits behind the same execution contract, so Wake's routing,
|
|
142
142
|
lifecycle, and sandbox behavior stay the same regardless of which CLI executes
|
|
143
|
-
a given
|
|
144
|
-
control plane itself. See [docs/runner-comparison.md](docs/runner-comparison.md)
|
|
143
|
+
a given activity. A fake runner adapter also exists for zero-token testing of
|
|
144
|
+
the control plane itself. See [docs/runner-comparison.md](docs/runner-comparison.md)
|
|
145
145
|
for capability differences between runners.
|
|
146
146
|
|
|
147
147
|
## Getting Started
|
|
@@ -228,26 +228,28 @@ and either:
|
|
|
228
228
|
|
|
229
229
|
## 2. External integrations
|
|
230
230
|
|
|
231
|
-
Ask the user
|
|
232
|
-
tracker, for example), and whether it should start polling immediately or
|
|
233
|
-
stay off until they're ready.
|
|
231
|
+
Ask the user:
|
|
234
232
|
|
|
235
|
-
|
|
233
|
+
- Which external source or sources should feed Wake work items?
|
|
234
|
+
- Which provider-specific rule should opt work in?
|
|
235
|
+
- Should polling start immediately, or stay disabled until they are ready?
|
|
236
|
+
|
|
237
|
+
Configure each provider under \`integrations\` in \`config.yaml\`:
|
|
236
238
|
|
|
237
239
|
\`\`\`yaml
|
|
238
240
|
integrations:
|
|
239
241
|
<name>:
|
|
240
242
|
provider: <provider-id>
|
|
241
243
|
enabled: true # or leave false to configure now, enable later
|
|
242
|
-
#
|
|
243
|
-
#
|
|
244
|
-
# fields it needs (which repositories/projects to watch, credentials,
|
|
245
|
-
# which events opt an item in).
|
|
244
|
+
# Provider-specific fields select resources, configure credentials,
|
|
245
|
+
# and define which observed work is admitted.
|
|
246
246
|
\`\`\`
|
|
247
247
|
|
|
248
|
-
|
|
249
|
-
|
|
250
|
-
configuration
|
|
248
|
+
The fields beneath an integration entry are provider-owned and validated by
|
|
249
|
+
that provider. Consult
|
|
250
|
+
https://github.com/atolis-hq/wake/blob/main/docs/configuration.md for each
|
|
251
|
+
supported provider's current schema, credential guidance, polling controls,
|
|
252
|
+
and intake rules.
|
|
251
253
|
|
|
252
254
|
## 3. Credential mounts (check before asking)
|
|
253
255
|
|
package/package.json
CHANGED
package/templates/SETUP.md
DELETED
|
@@ -1,115 +0,0 @@
|
|
|
1
|
-
?# Wake Setup Guide (for the assisting agent)
|
|
2
|
-
|
|
3
|
-
You are reading this because a human just ran `wake init` and asked you to
|
|
4
|
-
help finish configuring this Wake home. This file is written as instructions
|
|
5
|
-
to you, the assisting agent — not as prose for a human to read top to bottom.
|
|
6
|
-
|
|
7
|
-
Read `config.yaml` and `config.workflows.yaml` in this directory now — both
|
|
8
|
-
already exist with working defaults from `wake init`. Everything below tells
|
|
9
|
-
you which fields in those two files to change. Edit them directly; don't
|
|
10
|
-
create a new `config.<label>.yaml` split unless the user asks for one.
|
|
11
|
-
|
|
12
|
-
Work through the three sections below in order, asking the user only what's
|
|
13
|
-
asked in each section. Each section links the relevant part of the hosted
|
|
14
|
-
`docs/configuration.md` for full field-by-field reference beyond what's
|
|
15
|
-
summarized here.
|
|
16
|
-
|
|
17
|
-
## 1. GitHub source
|
|
18
|
-
|
|
19
|
-
Ask the user:
|
|
20
|
-
|
|
21
|
-
- Which GitHub repo(s) should Wake monitor for issues? (`owner/repo` format)
|
|
22
|
-
- Should polling start immediately, or stay off until they're ready?
|
|
23
|
-
|
|
24
|
-
Edit in `config.yaml`:
|
|
25
|
-
|
|
26
|
-
```yaml
|
|
27
|
-
sources:
|
|
28
|
-
github:
|
|
29
|
-
enabled: true # or leave false to configure now, enable later
|
|
30
|
-
repos: [owner/repo] # one or more, owner/repo format
|
|
31
|
-
```
|
|
32
|
-
|
|
33
|
-
Tell the user: once enabled, Wake adds/removes a
|
|
34
|
-
`wake:status.pending|working|failed|completed` label on issues it works, and
|
|
35
|
-
preserves any other labels already on the issue.
|
|
36
|
-
|
|
37
|
-
Full reference:
|
|
38
|
-
https://github.com/atolis-hq/wake/blob/main/docs/configuration.md#sourcesgithub
|
|
39
|
-
|
|
40
|
-
## 2. Runner and runnerPool
|
|
41
|
-
|
|
42
|
-
Ask the user which agent CLI(s) they have authenticated on this host:
|
|
43
|
-
Claude, Codex, and/or Cursor.
|
|
44
|
-
|
|
45
|
-
`config.workflows.yaml` already has example `runners` entries for
|
|
46
|
-
`claude-haiku`, `claude-opus`, `codex-mini`, `codex-flagship`, and
|
|
47
|
-
`cursor-composer`, but every runnerPool (`light`/`standard`/`deep`, with
|
|
48
|
-
`defaultRunnerPool: standard`) still points at the placeholder `fake` runner — none
|
|
49
|
-
of them route to a real runner yet. Don't rewrite this from scratch — pick
|
|
50
|
-
which runner(s) the user actually has access to, and either:
|
|
51
|
-
|
|
52
|
-
- repoint `runnerPools` so each runnerPool lists the real named runner(s) the user can
|
|
53
|
-
actually use instead of `fake`, or
|
|
54
|
-
- if the user has a runner not already listed (a different model, a
|
|
55
|
-
different CLI), add a new named entry under `runners` following the
|
|
56
|
-
existing pattern, then reference it from `runnerPools`.
|
|
57
|
-
- remove entries which are not needed.
|
|
58
|
-
|
|
59
|
-
Full reference:
|
|
60
|
-
https://github.com/atolis-hq/wake/blob/main/docs/configuration.md#runners
|
|
61
|
-
and
|
|
62
|
-
https://github.com/atolis-hq/wake/blob/main/docs/configuration.md#runnerPools
|
|
63
|
-
|
|
64
|
-
## 3. Credential mounts (check before asking)
|
|
65
|
-
|
|
66
|
-
Do not start by asking the user where their credentials are. First check the
|
|
67
|
-
host filesystem yourself for the files below, matching whichever runner(s)
|
|
68
|
-
were chosen in step 2:
|
|
69
|
-
|
|
70
|
-
- Claude: `~/.claude/.credentials.json` and `~/.claude/settings.json`
|
|
71
|
-
- Codex: `~/.codex/config.toml` and `~/.codex/auth.json`
|
|
72
|
-
- Cursor: `~/.config/cursor/auth.json`
|
|
73
|
-
|
|
74
|
-
For each file that exists, propose adding it to `sandbox.extraMounts` in
|
|
75
|
-
`config.yaml`, for example:
|
|
76
|
-
|
|
77
|
-
```yaml
|
|
78
|
-
sandbox:
|
|
79
|
-
extraMounts:
|
|
80
|
-
- source: /home/alice/.claude/.credentials.json
|
|
81
|
-
target: /home/wake/.claude/.credentials.json
|
|
82
|
-
readOnly: true
|
|
83
|
-
- source: /home/alice/.claude/settings.json
|
|
84
|
-
target: /home/wake/.claude/settings.json
|
|
85
|
-
readOnly: false
|
|
86
|
-
```
|
|
87
|
-
|
|
88
|
-
`.credentials.json`/`auth.json` should be `readOnly: true` unless the user
|
|
89
|
-
wants the sandbox able to refresh tokens on the host's behalf. `settings.json`
|
|
90
|
-
must stay `readOnly: false` — Claude plugin commands write to it. Use the
|
|
91
|
-
actual host home directory path (resolve `~` yourself; don't write a literal
|
|
92
|
-
tilde into YAML).
|
|
93
|
-
|
|
94
|
-
Never mount the whole `~/.claude`, `~/.codex`, or `~/.cursor` directory —
|
|
95
|
-
only the specific files listed above. Mounting the whole directory leaks
|
|
96
|
-
OS-specific absolute paths (e.g. Windows plugin cache paths) into the Linux
|
|
97
|
-
sandbox and can cause the sandbox to overwrite the host's plugin bookkeeping.
|
|
98
|
-
|
|
99
|
-
Only if none of the expected files exist for the runner the user chose, ask
|
|
100
|
-
them directly where their credentials live (e.g. a custom `CODEX_HOME`).
|
|
101
|
-
|
|
102
|
-
Full reference:
|
|
103
|
-
https://github.com/atolis-hq/wake/blob/main/docs/configuration.md#sandbox
|
|
104
|
-
|
|
105
|
-
## After config looks right
|
|
106
|
-
|
|
107
|
-
Don't try to explain the sandbox lifecycle yourself — point the user at (or
|
|
108
|
-
fetch, if you have web access):
|
|
109
|
-
|
|
110
|
-
- https://github.com/atolis-hq/wake/blob/main/docs/getting-started.md —
|
|
111
|
-
`wake sandbox build` / `up` / `setup` / `exec` / `down`
|
|
112
|
-
- https://github.com/atolis-hq/wake/blob/main/docs/runner-comparison.md —
|
|
113
|
-
deeper comparison of runner tradeoffs if the user asks which to pick
|
|
114
|
-
- https://github.com/atolis-hq/wake/blob/main/docs/configuration.md — every
|
|
115
|
-
config field, if something here doesn't cover their situation
|