@awebai/oats 0.69.22 → 0.69.24
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/bin/oats.mjs +185 -38
- package/docs/capabilities.md +365 -7
- package/docs/capability-manifest.schema.json +37 -1
- package/docs/configuration.md +27 -7
- package/docs/container.md +206 -228
- package/docs/design/2026-11-19-sandboxes.md +1352 -0
- package/docs/design/2026-11-20-provider-home-lifecycle.md +320 -0
- package/docs/design/README.md +7 -0
- package/docs/desktop-cli-api.md +673 -81
- package/docs/desktop.md +241 -65
- package/docs/execution-targets.md +186 -3
- package/docs/first-team.md +11 -0
- package/docs/implementation.md +6 -2
- package/docs/integrations.md +3 -3
- package/docs/knowledge-capability-authoring.md +10 -7
- package/docs/knowledge-reference/acceptance.md +21 -19
- package/docs/knowledge-reference/adoption.md +18 -4
- package/docs/knowledge-reference/model.md +11 -9
- package/docs/knowledge-reference/package-craft.md +2 -2
- package/docs/knowledge-reference/proposal.md +127 -0
- package/docs/knowledge-reference/provider-mapping.md +10 -10
- package/docs/knowledge-reference/reader-capture.md +26 -22
- package/docs/knowledge-theory.md +1 -1
- package/docs/knowledge.md +25 -16
- package/docs/official-catalog.md +4 -4
- package/docs/packages.md +13 -13
- package/docs/release-notes/v0.69.23.md +231 -0
- package/docs/release-notes/v0.69.24.md +264 -0
- package/docs/servers.md +10 -6
- package/docs/souls-and-instances.md +96 -22
- package/docs/tui.md +25 -4
- package/docs/upgrading.md +19 -5
- package/docs/workspaces.md +14 -6
- package/lib/argv.mjs +6 -2
- package/lib/capability-contract.mjs +2 -2
- package/lib/capability-show.mjs +11 -1
- package/lib/command-flags.mjs +40 -3
- package/lib/compile-cache.mjs +1 -1
- package/lib/core.mjs +1201 -195
- package/lib/errors.mjs +4 -1
- package/lib/hook-env.mjs +106 -0
- package/lib/host-run.mjs +11 -3
- package/lib/instance-board.mjs +337 -0
- package/lib/instance-events.mjs +48 -10
- package/lib/instance-inspect.mjs +44 -0
- package/lib/instance-lifecycle.mjs +34 -18
- package/lib/instance-listing.mjs +113 -16
- package/lib/instance-related-work.mjs +143 -0
- package/lib/kernel-features.mjs +13 -1
- package/lib/node-version.mjs +9 -0
- package/lib/provider-call-child.mjs +60 -0
- package/lib/provider-resume.mjs +83 -0
- package/lib/provider-test-seams.mjs +20 -0
- package/lib/read-call.mjs +2 -0
- package/lib/reserved-launch-env.mjs +8 -1
- package/lib/retire-admission.mjs +5 -1
- package/lib/schedule.mjs +14 -4
- package/lib/servers.mjs +151 -17
- package/lib/session-inspect.mjs +104 -13
- package/lib/session-providers.mjs +528 -0
- package/lib/session-resume.mjs +68 -15
- package/lib/stable-node.mjs +3 -0
- package/lib/triggers.mjs +3 -1
- package/package-catalog.json +3 -3
- package/package.json +2 -1
- package/packages/client/capability-list.mjs +21 -0
- package/packages/client/cli-adapter.mjs +19 -4
- package/packages/client/cli-locator.mjs +5 -1
- package/packages/client/deployment-data.mjs +27 -3
- package/packages/client/deployment-panel.mjs +27 -4
- package/packages/client/home-board.mjs +33 -19
- package/packages/client/inspect-boundary.mjs +18 -8
- package/packages/client/instance-acts.mjs +30 -3
- package/packages/client/instance-board.mjs +206 -0
- package/packages/client/instance-presentation.mjs +55 -0
- package/packages/client/instance-tree.mjs +34 -4
- package/packages/client/keymap.mjs +5 -0
- package/packages/client/liveness.mjs +15 -4
- package/packages/client/mail-contract.mjs +130 -0
- package/packages/client/own-codes.mjs +5 -1
- package/packages/client/panel-views.mjs +7 -7
- package/packages/client/related-work.mjs +84 -0
- package/packages/client/remote-roster.mjs +9 -3
- package/packages/client/session-backends.mjs +31 -0
- package/packages/client/session-start.mjs +18 -6
- package/packages/client/spawn-apply-cli.mjs +5 -3
- package/packages/client/spawn-apply-contract.mjs +25 -2
- package/packages/client/spawn-apply.mjs +5 -1
- package/packages/client/spawn-messages.mjs +3 -0
- package/packages/client/terminal-contract.mjs +4 -0
- package/packages/client/waiting-on-you.mjs +12 -2
- package/packages/client/workspace-cli.mjs +13 -2
- package/packages/tui/lib/model.mjs +14 -3
- package/packages/tui/lib/store.mjs +12 -3
- package/skills/oats-getting-started/SKILL.md +11 -8
- package/docs/knowledge-reference/harvester.md +0 -109
- package/lib/instance-same-work.mjs +0 -67
- package/packages/client/same-work.mjs +0 -49
package/docs/capabilities.md
CHANGED
|
@@ -333,8 +333,9 @@ A self-contained package has an `oats.json`:
|
|
|
333
333
|
- an exact name that is `.oats`, `.agents` or `.claude`, or that starts
|
|
334
334
|
with `.oats-events`, `.oats-stop`, `.oats-restart`, `.oats-rollback`,
|
|
335
335
|
`.oats-agents-md`, `.oats-start`, `.oats-attachments`, `.oats-color`,
|
|
336
|
-
`.oats-background`, `.oats-pinned
|
|
337
|
-
|
|
336
|
+
`.oats-background`, `.oats-pinned`, `.oats-related-work`, `.oats-same-work` (the
|
|
337
|
+
file of the flag `.oats-related-work` replaced, still read while the new one is absent) or
|
|
338
|
+
`.oats-board`: the top-level entries the kernel itself writes in an instance home;
|
|
338
339
|
- a prefix that starts with `.oats-`.
|
|
339
340
|
|
|
340
341
|
An exact name such as `.oats-aweb` is allowed: it is a provider's
|
|
@@ -440,6 +441,10 @@ A self-contained package has an `oats.json`:
|
|
|
440
441
|
- `triggerSources` (optional) declares the capability's
|
|
441
442
|
[trigger sources](#trigger-sources-triggersources). It is checked only
|
|
442
443
|
where a source is used, never when the capability loads.
|
|
444
|
+
- `sessionProviders` (optional) declares the session backends the
|
|
445
|
+
capability supplies beside the kernel's tmux
|
|
446
|
+
([session providers](#session-providers-sessionproviders)). It is checked
|
|
447
|
+
only where a provider is used, never when the capability loads.
|
|
443
448
|
- Target names never appear in a package manifest.
|
|
444
449
|
|
|
445
450
|
`capability` is the only manifest identity field; it may also carry
|
|
@@ -654,11 +659,17 @@ A spawn hook also gets `OATS_TASK`, `OATS_REPO`, `OATS_BRANCH`, `OATS_WORK`,
|
|
|
654
659
|
`OATS_TRIGGER_EVENT_FILE`. A launch hook also gets `OATS_HARNESS` and
|
|
655
660
|
`OATS_PREVIOUS_HARNESS`, and `OATS_LAUNCH_PREVIEW=1` when it runs for a
|
|
656
661
|
preview (only a preview-aware hook does, below); on a real run
|
|
657
|
-
`OATS_LAUNCH_PREVIEW` is not set. A
|
|
662
|
+
`OATS_LAUNCH_PREVIEW` is not set. A spawn hook and a launch hook also get
|
|
663
|
+
`OATS_BACKEND` (feature `session-providers`): the session
|
|
664
|
+
backend the instance launches on, `tmux` or the name of a
|
|
665
|
+
[session provider](#session-providers-sessionproviders). A spawn that launches
|
|
666
|
+
nothing (`--no-launch`) sets it too: it is what a later start launches on. A
|
|
667
|
+
kernel without the feature sets none, which means tmux. A worktree hook also gets the tree
|
|
658
668
|
variables of [The worktree event](#the-worktree-event).
|
|
659
669
|
|
|
660
670
|
A spawn or launch hook also gets `OATS_HARNESS_EXECUTABLE`, for every harness
|
|
661
|
-
and the same way in a preview
|
|
671
|
+
on tmux and the same way in a preview (a session provider runs the harness
|
|
672
|
+
itself, so the variable is absent there): the absolute real path of the executable the
|
|
662
673
|
pane will run, found the way the pane finds it. It is reserved, so a launch
|
|
663
674
|
configuration's `env` or a trigger cannot set it, and an inherited value is
|
|
664
675
|
dropped. It doesn't reach the harness's pane, even from an existing server's
|
|
@@ -1244,6 +1255,15 @@ bidi controls (U+202A–202E, U+2066–2069), U+200B, U+2060, U+FEFF and tag
|
|
|
1244
1255
|
characters (U+E0000–E007F) are refused, and everything else (emoji ZWJ
|
|
1245
1256
|
sequences, ZWNJ, LRM, RLM, ALM) is allowed.
|
|
1246
1257
|
|
|
1258
|
+
**The board's claim** (feature `instance-board`). An agent that needs a
|
|
1259
|
+
decision between 2 and 4 options asks it on its board, `oats instance board
|
|
1260
|
+
ask <question> --option <text> --option <text>`. While any board question is
|
|
1261
|
+
open, the kernel holds a claim of producer `oats.board`, reason `attention`,
|
|
1262
|
+
whose message is the oldest open question, and the last `oats instance board
|
|
1263
|
+
ask --clear <id>` removes it. It sits beside the agent's own `attention`
|
|
1264
|
+
claim and holds no scheduled wake. The board is in
|
|
1265
|
+
[desktop-cli-api.md](desktop-cli-api.md#the-agents-board-oats-instance-board).
|
|
1266
|
+
|
|
1247
1267
|
**The Claude Code emitter.** For a Claude instance, oats.core's spawn and
|
|
1248
1268
|
launch hooks (`bin/oats-core.mjs`, preview-aware) write Claude Code hooks
|
|
1249
1269
|
into the home's project settings, `<home>/.claude/settings.json`. Each one
|
|
@@ -1414,8 +1434,9 @@ waiting clear --producer oats.core --home <home>` from anywhere.
|
|
|
1414
1434
|
- **Its own entries only.** oats.core marks its entries by the absolute
|
|
1415
1435
|
path of its `claude-waiting.sh`. Each run removes only the entries that
|
|
1416
1436
|
name that path (and any matcher group or event array the removal
|
|
1417
|
-
empties), then appends its current ones.
|
|
1418
|
-
|
|
1437
|
+
empties), then appends its current ones. Apart from adding `tui` when
|
|
1438
|
+
absent (below), every other key and entry stays as it was, in order.
|
|
1439
|
+
The file is written atomically, mode 0600, and only
|
|
1419
1440
|
when its content changes. A temp file an interrupted write left behind
|
|
1420
1441
|
(`.claude/.settings.json.oats-core-<pid>-<ms>.tmp`) is removed by the next
|
|
1421
1442
|
spawn or start once its writer is gone, so it needs no retirement
|
|
@@ -1536,10 +1557,25 @@ that file replaces earlier ones wholesale, even one with no hooks. The
|
|
|
1536
1557
|
project `.claude/settings.json` composes with the user's settings (under any
|
|
1537
1558
|
`CLAUDE_CONFIG_DIR`) and with a `--settings`. So any capability that needs
|
|
1538
1559
|
Claude settings uses the same managed `<home>/.claude/settings.json` with
|
|
1539
|
-
its
|
|
1560
|
+
marked entries for its hooks, never `--settings`. oats.core leaves
|
|
1540
1561
|
`.claude/settings.local.json` to Claude Code, which writes its "don't ask
|
|
1541
1562
|
again" permission rules there.
|
|
1542
1563
|
|
|
1564
|
+
**Renderer selection.** In the same merge, at spawn and every launch,
|
|
1565
|
+
oats.core adds `"tui": "default"` when the project's settings have no `tui`
|
|
1566
|
+
key. Any existing value, including a non-string, is kept: a scalar has no
|
|
1567
|
+
marker to distinguish the person's choice from oats.core's. Any `tui` in
|
|
1568
|
+
Claude's merged settings skips its fullscreen-renderer offer, which can
|
|
1569
|
+
otherwise stop an unattended agent before it receives a wake. `"default"`
|
|
1570
|
+
selects the classic renderer used by the launch-prompt captures and keeps
|
|
1571
|
+
tmux scrollback. The cost: agent panes use Claude's classic renderer even
|
|
1572
|
+
where Claude defaults to fullscreen, and the project setting overrides a
|
|
1573
|
+
user-level `"tui": "fullscreen"`. An existing project `tui` still wins.
|
|
1574
|
+
A pass that changes nothing does not rewrite the file. The existing file
|
|
1575
|
+
refusals, preview behaviour and non-Claude behaviour above apply to `tui`
|
|
1576
|
+
too. If oats.core ever stops adding it, existing homes keep the key until
|
|
1577
|
+
respawn.
|
|
1578
|
+
|
|
1543
1579
|
**Limits.**
|
|
1544
1580
|
|
|
1545
1581
|
- If the user's Claude configuration sets `disableAllHooks` or
|
|
@@ -1934,3 +1970,325 @@ whole poll, resolution included, gets 35 s, and polls run only while they can
|
|
|
1934
1970
|
end within 50 s of the tick's start. Answer from a cache or a quick query: a
|
|
1935
1971
|
source slower than that should cache.
|
|
1936
1972
|
|
|
1973
|
+
## Session providers (`sessionProviders`)
|
|
1974
|
+
|
|
1975
|
+
A kernel that lists the feature `session-providers` in `oats version --json`
|
|
1976
|
+
has this contract. tmux is the kernel's own session backend. A capability can supply another one, a
|
|
1977
|
+
**session provider**: the instance's agent then runs wherever the provider
|
|
1978
|
+
runs it (a thread of another product, a remote runner), and the kernel reaches
|
|
1979
|
+
it through four of the capability's own commands. The kernel keeps what a
|
|
1980
|
+
backend must not own: the receipts and the authority check, the lifecycle
|
|
1981
|
+
claims, the refusals and the JSON answers. What an operator sees is in
|
|
1982
|
+
[execution-targets.md](execution-targets.md#session-providers).
|
|
1983
|
+
|
|
1984
|
+
```json
|
|
1985
|
+
"commands": { "session-launch": "bin/session.mjs launch", "session-inspect": "bin/session.mjs inspect",
|
|
1986
|
+
"session-input": "bin/session.mjs input", "session-stop": "bin/session.mjs stop",
|
|
1987
|
+
"session-inspect-many": "bin/session.mjs inspect-many" },
|
|
1988
|
+
"sessionProviders": {
|
|
1989
|
+
"bb": {
|
|
1990
|
+
"description": "Runs the agent as a bb thread",
|
|
1991
|
+
"harnesses": ["claude"],
|
|
1992
|
+
"commands": { "launch": "session-launch", "inspect": "session-inspect", "input": "session-input",
|
|
1993
|
+
"stop": "session-stop", "inspectMany": "session-inspect-many" }
|
|
1994
|
+
}
|
|
1995
|
+
}
|
|
1996
|
+
```
|
|
1997
|
+
|
|
1998
|
+
- The **name** is 1 to 40 lowercase letters, digits and dashes, starting with
|
|
1999
|
+
a letter or digit. `tmux`, `session-provider` and `herdr` are reserved. It
|
|
2000
|
+
is what `--backend` takes, and what a hook reads in `OATS_BACKEND`. At most
|
|
2001
|
+
16 providers.
|
|
2002
|
+
- **`commands`** names `launch`, `inspect`, `input` and `stop`, and optionally
|
|
2003
|
+
`inspectMany` and `check`. Each value is a key of the manifest's
|
|
2004
|
+
`commands`.
|
|
2005
|
+
- **`harnesses`** lists the harnesses the provider runs (1 to 16 names). A
|
|
2006
|
+
spawn or a start on another one is refused `E_BACKEND_HARNESS`.
|
|
2007
|
+
- **`description`** is optional: one line of at most 200 characters.
|
|
2008
|
+
- No other key.
|
|
2009
|
+
|
|
2010
|
+
**Containment.** A malformed `sessionProviders` never refuses the capability:
|
|
2011
|
+
it is validated where a provider is used. A problem inside one provider
|
|
2012
|
+
disables that provider only, and `--backend <name>` then answers
|
|
2013
|
+
`E_BACKEND_UNKNOWN` with the problem. `oats capabilities show <name> --json`
|
|
2014
|
+
adds `sessionProviders` (as declared) and, when there is one,
|
|
2015
|
+
`sessionProviderProblems: [{provider, pointer, message}]`.
|
|
2016
|
+
[`docs/capability-manifest.schema.json`](capability-manifest.schema.json)
|
|
2017
|
+
carries the same shape; that a verb names one of the manifest's commands is
|
|
2018
|
+
checked by the kernel only.
|
|
2019
|
+
|
|
2020
|
+
**One name, one meaning.** A provider is selected among the capabilities of
|
|
2021
|
+
the instance being spawned. Two of them declaring the same name is
|
|
2022
|
+
`E_BACKEND_AMBIGUOUS`: compose only one.
|
|
2023
|
+
|
|
2024
|
+
### The call
|
|
2025
|
+
|
|
2026
|
+
Every call runs the **home's own copy** of the capability:
|
|
2027
|
+
`node <home>/.oats/modules/<capability>/<script> <args>`. **Its working
|
|
2028
|
+
directory is not the home** (it is the filesystem root): read the home from
|
|
2029
|
+
the request (`input.home`) or `OATS_INSTANCE_HOME`, and leave
|
|
2030
|
+
nothing of the call's own running in the home (the session's processes work
|
|
2031
|
+
there; a helper of your command must not). OATS looks for
|
|
2032
|
+
processes working in a home before it resumes a lost session and before it
|
|
2033
|
+
retires the home, and a provider's call must never be one of them.
|
|
2034
|
+
|
|
2035
|
+
- **Environment:** what a lifecycle hook of the capability gets in that home
|
|
2036
|
+
([Commands and hooks](#commands-and-hooks)): `OATS_INSTANCE`,
|
|
2037
|
+
`OATS_INSTANCE_HOME`, `OATS_SETTINGS` (the settings captured at spawn),
|
|
2038
|
+
`OATS_META`, `OATS_CLI_BIN` and the rest, with `OATS_EVENT=session` and
|
|
2039
|
+
`OATS_BACKEND=<name>`.
|
|
2040
|
+
- **stdin:** one JSON request,
|
|
2041
|
+
`{ "schemaVersion": 1, "phase": "session", "capability": "<id>", "provider": "<name>", "verb": "<verb>", "input": { … } }`.
|
|
2042
|
+
- **stdout:** exactly one JSON document of at most 1 MiB, the same five header
|
|
2043
|
+
keys echoed, then `"ok": true, "result": { … }` or
|
|
2044
|
+
`"ok": false, "error": { "code": "E_…", "message": "…" }`. No other key.
|
|
2045
|
+
A refusal's code is `E_` followed by capitals, digits and underscores.
|
|
2046
|
+
- **Exit status 0**, for a result and for a refusal.
|
|
2047
|
+
- **Timeouts:** launch 120 s, stop 20 s plus the grace it is given, every
|
|
2048
|
+
other verb 20 s. The command is killed at its timeout.
|
|
2049
|
+
- **Process group:** OATS runs each command in a process group of its own,
|
|
2050
|
+
and ends the whole group (SIGKILL) at the timeout, and when the `oats`
|
|
2051
|
+
command that ran a `launch`, an `input` or a `stop` was itself killed
|
|
2052
|
+
while it waited (the next stop, start, input or retire of the home does
|
|
2053
|
+
it, before it calls you again). A stop also waits for an `input` that is
|
|
2054
|
+
being sent, up to its timeout, before it asks you to `stop`. **Run your helpers inside that group and never detach
|
|
2055
|
+
them** (no `setsid`, no daemon): a helper that escapes it can start a
|
|
2056
|
+
session OATS has no record of.
|
|
2057
|
+
- stderr is never read.
|
|
2058
|
+
|
|
2059
|
+
A call that does not end that way (another exit status, a signal, the
|
|
2060
|
+
timeout, no or malformed JSON, a wrong envelope, a result that breaks its
|
|
2061
|
+
verb's rules) is a **failed call**. A failed call is never read as an answer:
|
|
2062
|
+
an inspect that failed is never "stopped".
|
|
2063
|
+
|
|
2064
|
+
| Verb | A failed call, or a refusal the kernel does not relay |
|
|
2065
|
+
|---|---|
|
|
2066
|
+
| `inspect`, `inspectMany` | `E_SESSION_UNAVAILABLE`; the status row reads `running: null`, `runtimeState: "unreachable"` |
|
|
2067
|
+
| `check` | the readiness item reads `unknown`; a launch is refused `E_BACKEND_NOT_READY` |
|
|
2068
|
+
| `input` | `E_SESSION_INPUT_FAILED` |
|
|
2069
|
+
| `stop` | `E_SESSION_STOP_FAILED` |
|
|
2070
|
+
| `launch` | `E_SESSION_START_FAILED` |
|
|
2071
|
+
|
|
2072
|
+
The kernel relays these refusals under the provider's own code: from
|
|
2073
|
+
`launch`, any `E_BACKEND_*`; from `input`, `E_SESSION_PROMPT_OPEN` and
|
|
2074
|
+
`E_SESSION_INPUT_QUEUED`. The message of any refusal is carried into the
|
|
2075
|
+
kernel's, made safe to print and cut to 500 characters.
|
|
2076
|
+
|
|
2077
|
+
### launch
|
|
2078
|
+
|
|
2079
|
+
In:
|
|
2080
|
+
|
|
2081
|
+
```json
|
|
2082
|
+
{ "home": "/abs/home", "instance": "dev-a", "harness": "claude", "model": "id-or-null", "yolo": true,
|
|
2083
|
+
"env": [ { "name": "OATS_INSTANCE", "value": "dev-a", "source": "kernel" },
|
|
2084
|
+
{ "name": "AWEB_IDENTITY_HOME", "value": "/abs/…", "source": "capability", "capability": "oats.aweb" },
|
|
2085
|
+
{ "name": "REGION", "value": "eu", "source": "launch-config" },
|
|
2086
|
+
{ "name": "API_KEY", "value": "…", "source": "launch-config-reference", "secret": true } ],
|
|
2087
|
+
"unset": ["NAME"],
|
|
2088
|
+
"pathPrepend": ["/abs/home/.oats/bin"],
|
|
2089
|
+
"args": ["--flag", "value"],
|
|
2090
|
+
"promptFile": "/abs/home/TASK.md",
|
|
2091
|
+
"previous": null }
|
|
2092
|
+
```
|
|
2093
|
+
|
|
2094
|
+
- `env` is every variable the launch sets, in the order a tmux launch sets
|
|
2095
|
+
them. `source` says where each comes from: `kernel` (the instance's
|
|
2096
|
+
identity, a trigger's event file), `capability` (a hook's `env`, with the
|
|
2097
|
+
`capability` that owns it), `launch-config` (a launch configuration's
|
|
2098
|
+
literal value) or `launch-config-reference` (a launch configuration's
|
|
2099
|
+
`fromEnv` or `fromFile`: a secret source, marked `secret: true`, its value
|
|
2100
|
+
read on this host).
|
|
2101
|
+
- `unset` lists the names the launch removes: declared by a capability and
|
|
2102
|
+
left out of this launch.
|
|
2103
|
+
- `pathPrepend` is the home's kernel shim directory. Put it first on the
|
|
2104
|
+
agent's `PATH`, so its plain `oats` is the kernel that launched it.
|
|
2105
|
+
- `args` is the launch configuration's arguments, then the capabilities'
|
|
2106
|
+
launch arguments for the harness, as the words a shell would hand it
|
|
2107
|
+
(arguments the kernel cannot read as words are refused,
|
|
2108
|
+
`E_BACKEND_LAUNCH_ARG`, naming the capability). The kernel's own choices are never in
|
|
2109
|
+
it: they are `model` (null: the harness's default), `yolo` (`true`,
|
|
2110
|
+
`false`, or `null` when nobody chose) and the prompt.
|
|
2111
|
+
- `promptFile` is the home's `TASK.md`: the briefing a fresh conversation
|
|
2112
|
+
starts on.
|
|
2113
|
+
- `previous` is the endpoint of the session a start or a restart replaces;
|
|
2114
|
+
`null` for the home's first launch. The kernel has asked you to stop it
|
|
2115
|
+
just before, always (a restart's stop, once more for a home OATS had
|
|
2116
|
+
stopped, and for a session you answered as gone too: an input that reached
|
|
2117
|
+
it since may have had you restart it).
|
|
2118
|
+
|
|
2119
|
+
**A provider that cannot carry an env entry, an argument or `unset`
|
|
2120
|
+
refuses, naming it** (`E_BACKEND_LAUNCH_ENV`, `E_BACKEND_LAUNCH_ARG`). It
|
|
2121
|
+
never drops one silently, and it never stores a `secret` value where others
|
|
2122
|
+
can read it.
|
|
2123
|
+
|
|
2124
|
+
Out: `{ "endpoint": { … } }`. The endpoint is a JSON object whose canonical
|
|
2125
|
+
JSON is at most 4 KiB (4096 bytes). **It names one session: answer an
|
|
2126
|
+
endpoint you never answered before for that home** (a new thread id, not a
|
|
2127
|
+
locator a later session takes over). The kernel stops a session by its
|
|
2128
|
+
endpoint, also after a newer one was launched. The kernel records it as given, in the
|
|
2129
|
+
home's `instance.json` and its receipt, and hands it back on every call. It
|
|
2130
|
+
holds **locators only, never a credential**.
|
|
2131
|
+
|
|
2132
|
+
The kernel refuses, before anything is created, a launch configuration that
|
|
2133
|
+
declares `executable` for a provider backend (`E_BACKEND_LAUNCH_ARG`): the
|
|
2134
|
+
provider runs the harness itself.
|
|
2135
|
+
|
|
2136
|
+
### inspect and inspectMany
|
|
2137
|
+
|
|
2138
|
+
`inspect` in: `{ "home", "endpoint" }`. Out:
|
|
2139
|
+
`{ "present": true|false, "state": "unknown"|"shell"|"stopped", "running": true|false|null, "open"?: { "label", "url" } }`,
|
|
2140
|
+
exactly those keys.
|
|
2141
|
+
|
|
2142
|
+
| The session | present | state | running |
|
|
2143
|
+
|---|---|---|---|
|
|
2144
|
+
| a turn or a start is in progress | true | `unknown` | true |
|
|
2145
|
+
| there, and the provider cannot tell a live idle session from a settled one | true | `unknown` | null |
|
|
2146
|
+
| gone: missing, deleted, or in error with nothing running | false | `stopped` | false |
|
|
2147
|
+
| the harness ended and something else holds the session, as a terminal left at its shell | true | `shell` | false |
|
|
2148
|
+
|
|
2149
|
+
A session that is not there (`present: false`) is `stopped` and runs nothing:
|
|
2150
|
+
any other answer with `present: false` is outside the table, and is read as
|
|
2151
|
+
no answer (unavailable), like a word outside the contract.
|
|
2152
|
+
|
|
2153
|
+
Map only what the product shows; never infer. **A session that is gone on a
|
|
2154
|
+
home OATS did not stop is a lost session**: OATS resumes it through `input`,
|
|
2155
|
+
under its cap ([below](#what-the-kernel-does-around-the-verbs)). So answer a
|
|
2156
|
+
session whose run ended in error as gone, not as `shell`, when your product
|
|
2157
|
+
restarts it on the next input. `open` is what a client may
|
|
2158
|
+
show instead of a terminal: a one-line `label` of at most 100 characters and
|
|
2159
|
+
an `http` or `https` `url` of at most 2000, without credentials.
|
|
2160
|
+
|
|
2161
|
+
`inspectMany` (optional) in:
|
|
2162
|
+
`{ "homes": [ { "home", "instance", "endpoint" } ] }`, at most 200. Out:
|
|
2163
|
+
`{ "results": [ … ] }`, one per home in the same order, each an inspect
|
|
2164
|
+
result or `{ "error": { "code", "message" } }`. `oats status` calls it once
|
|
2165
|
+
per group of homes that share the capability's module copy (same digest),
|
|
2166
|
+
the provider and the settings; the call's environment is the group's first
|
|
2167
|
+
home's. Each entry is judged as one `inspect` answer would be. A batch that
|
|
2168
|
+
fails or is not one result per home, and an entry that is neither a result
|
|
2169
|
+
nor a refusal, fall back to that home's own `inspect`, never to "stopped".
|
|
2170
|
+
Without `inspectMany`, status calls `inspect` per home.
|
|
2171
|
+
|
|
2172
|
+
**The kernel never calls inspect for a home OATS stopped** (below).
|
|
2173
|
+
|
|
2174
|
+
### input
|
|
2175
|
+
|
|
2176
|
+
In: `{ "home", "endpoint", "text" }`. Out: `{ "submitted": true, "verified": true|false }`
|
|
2177
|
+
**only when the text was dispatched** to the session. `submitted` is never
|
|
2178
|
+
`false` in a result: an input that was not dispatched is a refusal.
|
|
2179
|
+
|
|
2180
|
+
- A session with a prompt open (a permission prompt, a question) is not
|
|
2181
|
+
answered by an input: refuse `E_SESSION_PROMPT_OPEN` and send nothing.
|
|
2182
|
+
Otherwise any mail could approve a permission.
|
|
2183
|
+
- An input the product queued instead of dispatching is not a submission:
|
|
2184
|
+
withdraw it and refuse `E_SESSION_INPUT_QUEUED`.
|
|
2185
|
+
|
|
2186
|
+
Both reach the caller as they are, with `ok: false`. The messaging broker
|
|
2187
|
+
leaves such a mail unread; a scheduled wake records it not delivered.
|
|
2188
|
+
|
|
2189
|
+
### stop
|
|
2190
|
+
|
|
2191
|
+
In: `{ "home", "endpoint", "graceSeconds", "launchSent"? }`. Out: `{ "stopped": true }`.
|
|
2192
|
+
`endpoint` is `null` when the kernel has no endpoint for a session you may
|
|
2193
|
+
have started: a launch whose answer was lost, and a launch it sent whose
|
|
2194
|
+
spawn or start ended before recording your answer (it was killed while you
|
|
2195
|
+
launched). Then stop whatever live session is rooted at the home, and answer
|
|
2196
|
+
`stopped: true` also when there is none. The input then also carries
|
|
2197
|
+
`"launchSent": "<instant>"`, the time the kernel recorded for that launch,
|
|
2198
|
+
when it has one: what the launch started was created at or after it, so you
|
|
2199
|
+
can leave alone what was in the home's place before (a person's own session
|
|
2200
|
+
there). **The kernel relies on this stop**: it is the only way a session
|
|
2201
|
+
started by a killed spawn is ever stopped.
|
|
2202
|
+
|
|
2203
|
+
### check (optional)
|
|
2204
|
+
|
|
2205
|
+
The provider's readiness. In: `{ "home": "/abs/home" | null, "soul": "<name>" | null, "settings": { … } }`:
|
|
2206
|
+
a home, or, before any home exists, a soul. Out:
|
|
2207
|
+
`{ "status": "ready" | "needs-configuration" | "authorization-required" | "unavailable", "problems": [ { "code", "message" } ], "warnings": [ { "code", "message" } ] }`
|
|
2208
|
+
(`warnings` optional; no problem on `ready`). Say where a credential comes
|
|
2209
|
+
from, never print it.
|
|
2210
|
+
|
|
2211
|
+
- `oats readiness` relays the answer as a `providers` item, the shape a
|
|
2212
|
+
binding check's answer has: for a home, its own backend's, required; for a
|
|
2213
|
+
soul, each provider its capabilities offer, not required (the soul can
|
|
2214
|
+
launch on tmux).
|
|
2215
|
+
- **The kernel asks it right before every launch** (a spawn's, a start's, a
|
|
2216
|
+
restart's before its stop). A status other than `ready` refuses the launch:
|
|
2217
|
+
with the first problem's code when that is an `E_BACKEND_*` code, else
|
|
2218
|
+
`E_BACKEND_NOT_READY`, the message carrying your problems. A check that
|
|
2219
|
+
fails or answers outside these words is not ready.
|
|
2220
|
+
|
|
2221
|
+
### Requirements by backend (`requires[].backends`)
|
|
2222
|
+
|
|
2223
|
+
A harness-package requirement (`requires: [{ harness, package, … }]`) is for
|
|
2224
|
+
the harness install on this host. A row may say which backends it is for:
|
|
2225
|
+
`"backends": ["tmux"]`, `["session-provider"]` (any session provider) or
|
|
2226
|
+
both. Without the key it applies to every backend, a provider's launch
|
|
2227
|
+
included, where it is probed on this host. It is a key of its own, never
|
|
2228
|
+
inside `when`, whose keys are the capability's settings. A kernel without
|
|
2229
|
+
`session-providers` knows no such key and applies the row to every launch:
|
|
2230
|
+
a package that uses `backends` raises its OATS floor to a kernel that lists
|
|
2231
|
+
the feature.
|
|
2232
|
+
|
|
2233
|
+
### What the kernel does around the verbs
|
|
2234
|
+
|
|
2235
|
+
- **Stop stays stopped.** At every OATS stop (`oats instance stop`, a
|
|
2236
|
+
restart's stop, a retire, `oats host run`'s shutdown) the kernel records
|
|
2237
|
+
`stoppedAt` in the home's receipt **before** it calls `stop`, and a start
|
|
2238
|
+
clears it. While it is set, inspect answers stopped (`present: false`,
|
|
2239
|
+
`running: false`) without calling the provider, and every input (a
|
|
2240
|
+
person's `oats session input`, a scheduled wake, a client, the messaging
|
|
2241
|
+
broker) is refused `E_SESSION_NOT_RUNNING` before any call. A provider
|
|
2242
|
+
whose product restarts a session on its next input can rely on this.
|
|
2243
|
+
- **Hooks first.** The capabilities' spawn hooks (at a spawn) and launch
|
|
2244
|
+
hooks (at a start or a restart) run before `launch`, as before a tmux
|
|
2245
|
+
window, with `OATS_BACKEND`.
|
|
2246
|
+
- **Retire** records `stoppedAt`, calls `stop`, then runs the retire hooks
|
|
2247
|
+
and removes the home. The retire hook is not handed the endpoint: a
|
|
2248
|
+
provider that archives its session finds it by the home.
|
|
2249
|
+
- **A launch that fails** leaves the home as it was (stopped, or never
|
|
2250
|
+
launched; a spawn is rolled back). When its answer was lost, the kernel
|
|
2251
|
+
calls `stop` with `endpoint: null` first.
|
|
2252
|
+
- **A launch that was sent and never recorded.** The kernel writes in the
|
|
2253
|
+
home's receipt that a launch is being sent, right before it calls you, and
|
|
2254
|
+
takes that away when it has recorded your endpoint. A spawn or a start
|
|
2255
|
+
that is killed while you launch leaves it there: the home then reads as
|
|
2256
|
+
unknown, and the next stop, start or retire first calls `stop` with
|
|
2257
|
+
`endpoint: null`.
|
|
2258
|
+
- **Which stop it was** is on the home's stop receipt, never a field of
|
|
2259
|
+
yours: `oats host run`'s shutdown is a loss that the sweep resumes, an
|
|
2260
|
+
operator's stop is the intent that keeps the home down.
|
|
2261
|
+
- **A lost session is resumed through `input`.** A home whose session you
|
|
2262
|
+
answer as gone, and that OATS did not stop, is resumed by the sweep with
|
|
2263
|
+
one notice, and any other input to it is a resume too. The kernel counts
|
|
2264
|
+
each against the home's cap before it calls `input` (at most 3 in an hour,
|
|
2265
|
+
within one boot), refuses at the cap with `E_SESSION_NOT_RUNNING` without
|
|
2266
|
+
calling you, and first waits (a resume) or refuses (an input) while a
|
|
2267
|
+
process still works in the home. So an `inspect` precedes every `input` to
|
|
2268
|
+
a home OATS did not stop (the resume of a home stopped with the host sends
|
|
2269
|
+
its notice without one).
|
|
2270
|
+
An input you refuse (`ok: false`: nothing was sent) is not counted. One
|
|
2271
|
+
whose call fails or answers outside the rules stays counted: you may have
|
|
2272
|
+
restarted the session on it.
|
|
2273
|
+
- **`stop` is idempotent.** The kernel calls it again wherever it needs the
|
|
2274
|
+
session gone: an operator's stop of a home that is already stopped (how a
|
|
2275
|
+
stop you did not confirm is finished), a start of a stopped home before
|
|
2276
|
+
its launch, and after an input that crossed a stop (an input holds no
|
|
2277
|
+
claim: when a stop or a restart lands while one is being sent, the kernel
|
|
2278
|
+
asks you to stop that session again and does not deliver the input,
|
|
2279
|
+
whatever you answered to it, a refusal included). After a restart it asks
|
|
2280
|
+
that only when the new session has another endpoint: where you hand a
|
|
2281
|
+
restart the same locator again, that locator is the new session's, and the
|
|
2282
|
+
kernel tells the two apart by its own record of the start. An input that
|
|
2283
|
+
is in flight when a start begins its launch is such a crossing too.
|
|
2284
|
+
Answer `stopped: true` for a session that is already stopped.
|
|
2285
|
+
- **A spawn holds the home while you launch.** From before the home can be
|
|
2286
|
+
acted on until your launch's endpoint is recorded (or the spawn is undone),
|
|
2287
|
+
the spawn holds the home's lifecycle claim: a stop, a start or a retire
|
|
2288
|
+
that arrives meanwhile is refused as busy, never answered as if there were
|
|
2289
|
+
nothing to stop.
|
|
2290
|
+
- **Nothing of your endpoint leaves the host.** A status row carries the
|
|
2291
|
+
capability and, on its own host only, the endpoint's `threadId` when it
|
|
2292
|
+
is a string; the endpoint is on local inspect, spawn and start answers.
|
|
2293
|
+
`open` is relayed: make its `url` one a viewer may open, never your API's
|
|
2294
|
+
address off its host, and put no token in either.
|
|
@@ -165,6 +165,13 @@
|
|
|
165
165
|
"description": "Applies only when every named capability setting has the given effective value (a requirement conditional on configuration, e.g. delivery: channel).",
|
|
166
166
|
"additionalProperties": {}
|
|
167
167
|
},
|
|
168
|
+
"backends": {
|
|
169
|
+
"type": "array",
|
|
170
|
+
"description": "The session backends this requirement is for (feature session-providers): \"tmux\", \"session-provider\" (any session provider), or both. Without the key the requirement applies to every backend. A manifest that uses it needs a kernel that lists session-providers: an older kernel applies the row to every launch.",
|
|
171
|
+
"minItems": 1,
|
|
172
|
+
"uniqueItems": true,
|
|
173
|
+
"items": { "enum": ["tmux", "session-provider"] }
|
|
174
|
+
},
|
|
168
175
|
"minVersion": {
|
|
169
176
|
"type": "string",
|
|
170
177
|
"description": "Lowest acceptable installed version of the package, read from the package.json under the install directory the harness's listing names; an older or absent manifest fails the requirement with the install remedy.",
|
|
@@ -248,6 +255,35 @@
|
|
|
248
255
|
}
|
|
249
256
|
}
|
|
250
257
|
},
|
|
258
|
+
"sessionProviders": {
|
|
259
|
+
"description": "Session providers (feature session-providers): session backends this capability supplies beside the kernel's own tmux. `oats spawn --backend <name>` (a schedule's backend, a trigger's spawn.backend) selects one among the instance's capabilities; the kernel then calls the named commands for that instance's session, each with one JSON request on stdin and one JSON answer on stdout (docs/capabilities.md, Session providers). Validated only where a provider is used, never when the capability loads: a malformed declaration cannot be selected (E_BACKEND_UNKNOWN) and nothing else changes. What JSON Schema cannot say is checked there too: each command names one of this manifest's commands.",
|
|
260
|
+
"type": "object",
|
|
261
|
+
"maxProperties": 16,
|
|
262
|
+
"propertyNames": { "pattern": "^[a-z0-9][a-z0-9-]{0,39}$", "not": { "enum": ["tmux", "session-provider", "herdr"] } },
|
|
263
|
+
"additionalProperties": {
|
|
264
|
+
"type": "object",
|
|
265
|
+
"required": ["commands", "harnesses"],
|
|
266
|
+
"additionalProperties": false,
|
|
267
|
+
"properties": {
|
|
268
|
+
"description": { "type": "string", "minLength": 1, "maxLength": 200, "pattern": "^[^\\u0000-\\u001f\\u007f-\\u009f\\u2028\\u2029]*$" },
|
|
269
|
+
"harnesses": { "type": "array", "minItems": 1, "maxItems": 16, "uniqueItems": true, "items": { "type": "string", "pattern": "^[a-z][a-z0-9-]{0,39}$" }, "description": "The harnesses this backend runs. A spawn or a start on another is refused E_BACKEND_HARNESS before anything is created or stopped." },
|
|
270
|
+
"commands": {
|
|
271
|
+
"type": "object",
|
|
272
|
+
"required": ["launch", "inspect", "input", "stop"],
|
|
273
|
+
"additionalProperties": false,
|
|
274
|
+
"description": "The provider's verbs, each naming one of this manifest's commands.",
|
|
275
|
+
"properties": {
|
|
276
|
+
"launch": { "type": "string", "minLength": 1, "description": "Starts the instance's session; answers the endpoint the kernel records." },
|
|
277
|
+
"inspect": { "type": "string", "minLength": 1, "description": "Answers one session's { present, state, running, open? }." },
|
|
278
|
+
"input": { "type": "string", "minLength": 1, "description": "Sends one text to the session; answers { submitted: true, verified } only when it was dispatched." },
|
|
279
|
+
"stop": { "type": "string", "minLength": 1, "description": "Stops the session; answers { stopped: true }." },
|
|
280
|
+
"inspectMany": { "type": "string", "minLength": 1, "description": "Optional: inspects several homes' sessions in one call (oats status)." },
|
|
281
|
+
"check": { "type": "string", "minLength": 1, "description": "Optional: the provider's readiness for a home or a soul, { status, problems, warnings }. oats readiness relays it, and the kernel asks it right before every launch." }
|
|
282
|
+
}
|
|
283
|
+
}
|
|
284
|
+
}
|
|
285
|
+
}
|
|
286
|
+
},
|
|
251
287
|
"triggerSources": {
|
|
252
288
|
"description": "Trigger sources (#669): each a command of this capability that, at each poll, lists the events due now; a trigger names one as on.source \"<capability>:<source>\". Validated only where a source is used (trigger add, a poll, trigger test and trigger poll), never when the capability loads: a malformed declaration fails the triggers naming it (E_TRIGGER_SOURCE) and nothing else. Checked at use only, not by this schema: every pattern compiles; command names a key of commands; a parameter default matches its pattern. A pattern is ECMA-262 with the u flag, matched anchored as ^(?:…)$; the default parameter value pattern is ^[A-Za-z0-9._/:@ ,*-]{0,200}$ and the default field pattern ^[A-Za-z0-9._/:@-]{1,200}$. Whatever its pattern admits, an event's field value is at most 200 code points and carries no control, bidi, invisible or tag character.",
|
|
253
289
|
"type": "object",
|
|
@@ -374,7 +410,7 @@
|
|
|
374
410
|
"description": "Home entries the kernel owns cannot be declared.",
|
|
375
411
|
"anyOf": [
|
|
376
412
|
{"pattern": "^\\.(?:oats|agents|claude)$"},
|
|
377
|
-
{"pattern": "^\\.oats-(?:events|stop|restart|rollback|agents-md|start|attachments|color|background|pinned|same-work)"},
|
|
413
|
+
{"pattern": "^\\.oats-(?:events|stop|restart|rollback|agents-md|start|attachments|color|background|pinned|same-work|related-work|board)"},
|
|
378
414
|
{"pattern": "^\\.oats-.*\\*$"}
|
|
379
415
|
]
|
|
380
416
|
}
|
package/docs/configuration.md
CHANGED
|
@@ -40,7 +40,6 @@ launch-configs: # named ways this host starts a har
|
|
|
40
40
|
personal:
|
|
41
41
|
harness: claude
|
|
42
42
|
executable: "./bin/claude-wrapper.sh" # relative → against this deployment directory
|
|
43
|
-
args: ["--verbose"]
|
|
44
43
|
env:
|
|
45
44
|
CLAUDE_CONFIG_DIR: { fromEnv: PERSONAL_CLAUDE_DIR }
|
|
46
45
|
model: opus
|
|
@@ -304,6 +303,19 @@ configuration is a host choice: a soul never names one.
|
|
|
304
303
|
crosses harnesses.
|
|
305
304
|
- The executable must be a regular executable file; it is never run to probe
|
|
306
305
|
it.
|
|
306
|
+
- **`args` and package verification.** When any active capability declares
|
|
307
|
+
an applicable harness package requirement for the selected harness, a
|
|
308
|
+
configuration with any non-empty `args` is refused at spawn with
|
|
309
|
+
`E_LAUNCH_PROBE_UNSUPPORTED`; nothing is created. With `oats.aweb`, every
|
|
310
|
+
Claude launch has such a requirement. The package probe runs
|
|
311
|
+
`claude plugin list --json`, which cannot carry the launch's args;
|
|
312
|
+
arguments such as `--settings` can change which plugins load, so that
|
|
313
|
+
probe could not verify the actual launch. There is no passthrough or
|
|
314
|
+
exception for permission flags. Put native settings in the user
|
|
315
|
+
`settings.json` of the Claude config the launch uses: for agents only,
|
|
316
|
+
set `CLAUDE_CONFIG_DIR` through the configuration's `env`, which the
|
|
317
|
+
probe does carry. The other supported route is a wrapper `executable`,
|
|
318
|
+
which both the launch and the package probe use.
|
|
307
319
|
- `oats launch-config list` shows the effective entries;
|
|
308
320
|
`oats launch-config set <name> --file <json>` and
|
|
309
321
|
`oats launch-config remove <name>` rewrite only this block, under the
|
|
@@ -315,9 +327,10 @@ configuration is a host choice: a soul never names one.
|
|
|
315
327
|
`deprecated-runtime-name` warning.
|
|
316
328
|
- **Claude's permissions** are Claude's own settings, not OATS keys. They go
|
|
317
329
|
in the user settings of the Claude config a launch uses (its
|
|
318
|
-
`CLAUDE_CONFIG_DIR`, else `~/.claude`)
|
|
319
|
-
|
|
320
|
-
capabilities manage it at every start,
|
|
330
|
+
`CLAUDE_CONFIG_DIR`, else `~/.claude`): set
|
|
331
|
+
`"permissions": {"defaultMode": "auto"}` there for auto mode. Never in a
|
|
332
|
+
home's `.claude/settings.json`: the capabilities manage it at every start,
|
|
333
|
+
and Claude ignores auto mode there.
|
|
321
334
|
oats.setup's `/oats-agent-permissions` (oats.framework 1.13.0) teaches the
|
|
322
335
|
setup: auto mode, narrow allow rules for the `oats` and `gh` reads,
|
|
323
336
|
`oats spawn` and reading `aw` mail and chat, and what stays gated.
|
|
@@ -334,10 +347,14 @@ person; nothing in OATS writes it.
|
|
|
334
347
|
launch-configs:
|
|
335
348
|
claude-auto:
|
|
336
349
|
harness: claude
|
|
337
|
-
|
|
350
|
+
env:
|
|
351
|
+
CLAUDE_CONFIG_DIR: { fromEnv: AGENTS_CLAUDE_DIR }
|
|
338
352
|
default: true
|
|
339
353
|
```
|
|
340
354
|
|
|
355
|
+
Here `AGENTS_CLAUDE_DIR` names an agents-only Claude config directory;
|
|
356
|
+
its user `settings.json` sets `"permissions": {"defaultMode": "auto"}`.
|
|
357
|
+
|
|
341
358
|
- **One key.** The file holds `launch-configs:` and nothing else. Its entries
|
|
342
359
|
have exactly the schema of the deployment's block and the same checks.
|
|
343
360
|
- **Missing or broken.** A missing file declares nothing. A file that is there
|
|
@@ -385,7 +402,8 @@ launch-configs:
|
|
|
385
402
|
`default: true` makes a configuration this host's baseline for its harness
|
|
386
403
|
(0.32, feature `launch-config-default`). Use it for what every launch of a
|
|
387
404
|
harness on this machine needs, whatever soul or preference chose it: an
|
|
388
|
-
account directory (`CLAUDE_CONFIG_DIR`)
|
|
405
|
+
account directory (`CLAUDE_CONFIG_DIR`) or a wrapper `executable`. Its `args`
|
|
406
|
+
are subject to the package-verification refusal above.
|
|
389
407
|
|
|
390
408
|
- **When it applies:** a new launch that picks the harness without naming a
|
|
391
409
|
configuration: a soul's `launch:`, an inline `souls.launch` preference,
|
|
@@ -632,7 +650,9 @@ came from, the executable, args, env, model, yolo, and each capability's
|
|
|
632
650
|
launch contribution with its settings and trust, and the kernel that launched
|
|
633
651
|
it (`kernelBin`, re-written by every start). One renderer turns it into
|
|
634
652
|
the `command`. Configuration `args` go after the harness's own options and
|
|
635
|
-
before capability arguments; every argument is single-quoted.
|
|
653
|
+
before capability arguments; every argument is single-quoted. The
|
|
654
|
+
[package-verification refusal](#launch-configurations) applies to these
|
|
655
|
+
`args`.
|
|
636
656
|
|
|
637
657
|
**Starting and restarting a home.** `oats session start --home <abs>` runs
|
|
638
658
|
the recorded recipe. With `--launch-config`, `--harness` or `--yolo`, the
|