@ctlflow/fleet 0.0.0-stage → 0.1.1
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 +46 -2
- package/dist/bin/fleet.mjs +5919 -0
- package/docs/README.md +11 -0
- package/docs/commands.md +136 -0
- package/docs/configuration.md +239 -0
- package/docs/getting-started.md +111 -0
- package/docs/operations.md +85 -0
- package/docs/remote.md +46 -0
- package/docs/sessions.md +109 -0
- package/package.json +20 -4
package/README.md
CHANGED
|
@@ -1,3 +1,47 @@
|
|
|
1
|
-
#
|
|
1
|
+
# fleet
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
The CLI manages user-scoped fleetd controllers locally or through SSH. Every
|
|
4
|
+
Crossfire agent belongs to a fleet, including a one-agent project. fleetd owns
|
|
5
|
+
the shared listener and native sessions; the CLI uses its private API. Crossfire
|
|
6
|
+
owns enrollment, room permissions, wake rules and listener activity.
|
|
7
|
+
|
|
8
|
+
## Guides
|
|
9
|
+
- [Guide index](docs/README.md)
|
|
10
|
+
- [Installation and first fleet](docs/getting-started.md)
|
|
11
|
+
- [TOML files and all runtime settings](docs/configuration.md)
|
|
12
|
+
- [Command reference and lifecycle](docs/commands.md)
|
|
13
|
+
- [Remote controllers and native terminal pointers](docs/remote.md)
|
|
14
|
+
- [Updates, backups and troubleshooting](docs/operations.md)
|
|
15
|
+
- [Controller API and ownership](https://github.com/control-flow-project/crossfire/blob/main/services/fleetd/README.md)
|
|
16
|
+
- [Crossfire management](https://github.com/control-flow-project/crossfire/blob/main/services/crossfire/docs/management.md)
|
|
17
|
+
|
|
18
|
+
## First setup
|
|
19
|
+
|
|
20
|
+
Open the personalized setup guide from Crossfire management. Its command installs
|
|
21
|
+
the exact public `@ctlflow/fleet` release in private user storage and invokes:
|
|
22
|
+
|
|
23
|
+
```sh
|
|
24
|
+
fleet setup --descriptor 'https://YOUR_SERVER/onboarding/setup.json?ownerMatrixId=ENCODED_OWNER_MATRIX_ID'
|
|
25
|
+
```
|
|
26
|
+
|
|
27
|
+
Run from the project directory. Setup asks no configuration questions. It starts
|
|
28
|
+
or reuses the private controller, persists stable names, requests real human
|
|
29
|
+
approval and verifies model-free listener readiness. The coding session doing
|
|
30
|
+
setup stays independent; it is not adopted. Native project files and its existing
|
|
31
|
+
`.codex` configuration/history are not overwritten. Optional
|
|
32
|
+
`--config /absolute/runtime.toml` uses the same ordinary fleet model.
|
|
33
|
+
|
|
34
|
+
Repeated setup reuses the controller, fleet, accounts and managed histories.
|
|
35
|
+
Explicitly stopped agents stay stopped. Normal `fleet start FLEET` resumes full
|
|
36
|
+
saved history; only `--fresh-history` requests replacement. Read-only commands
|
|
37
|
+
never start a controller or wake a native session. `fleet test FLEET` checks
|
|
38
|
+
actual listener/authentication/native readiness without a model turn.
|
|
39
|
+
|
|
40
|
+
Startup creates no chats. Private owner DMs open on first contact, visible output
|
|
41
|
+
or native permission request; additional room membership is explicit.
|
|
42
|
+
|
|
43
|
+
Codex defaults to `danger-full-access` and `on-request`. Sandbox, approval
|
|
44
|
+
policy, model and reasoning remain configurable in the definition's
|
|
45
|
+
`[runtime_options]`. The native-named
|
|
46
|
+
`dangerously_bypass_approvals_and_sandbox = true` selects full access without
|
|
47
|
+
approvals; it is never enabled implicitly.
|