owenloop 0.5.0 → 0.5.1

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.
Files changed (211) hide show
  1. package/CHANGELOG.md +45 -0
  2. package/README.md +137 -35
  3. package/bin/owenloop.mjs +49 -1
  4. package/dist/packages/work/src/agent/brief.d.ts +8 -4
  5. package/dist/packages/work/src/agent/brief.js +7 -3
  6. package/dist/packages/work/src/agent/loop.d.ts +13 -7
  7. package/dist/packages/work/src/agent/loop.js +136 -9
  8. package/dist/packages/work/src/agent/model-policy.d.ts +49 -0
  9. package/dist/packages/work/src/agent/model-policy.js +138 -0
  10. package/dist/packages/work/src/bundle/cache.d.ts +7 -5
  11. package/dist/packages/work/src/bundle/cache.js +20 -6
  12. package/dist/packages/work/src/bundle/fetch.d.ts +7 -5
  13. package/dist/packages/work/src/bundle/fetch.js +27 -7
  14. package/dist/packages/work/src/bundle/types.d.ts +8 -4
  15. package/dist/packages/work/src/consumed-verifier.d.ts +45 -0
  16. package/dist/packages/work/src/consumed-verifier.js +205 -0
  17. package/dist/packages/work/src/exec/instructions.d.ts +99 -0
  18. package/dist/packages/work/src/exec/instructions.js +323 -0
  19. package/dist/packages/work/src/exec/loop.d.ts +12 -1
  20. package/dist/packages/work/src/exec/loop.js +137 -20
  21. package/dist/packages/work/src/exec/payload.d.ts +29 -0
  22. package/dist/packages/work/src/exec/payload.js +54 -0
  23. package/dist/packages/work/src/exec/receipt.d.ts +9 -6
  24. package/dist/packages/work/src/exec/receipt.js +12 -1
  25. package/dist/packages/work/src/exec/runner.d.ts +7 -1
  26. package/dist/packages/work/src/exec/runner.js +74 -3
  27. package/dist/packages/work/src/harness/child-env.js +7 -0
  28. package/dist/packages/work/src/harness/claude.d.ts +20 -8
  29. package/dist/packages/work/src/harness/claude.js +260 -58
  30. package/dist/packages/work/src/harness/codex.d.ts +5 -4
  31. package/dist/packages/work/src/harness/codex.js +169 -41
  32. package/dist/packages/work/src/harness/contract.d.ts +36 -11
  33. package/dist/packages/work/src/harness/fake.js +1 -0
  34. package/dist/packages/work/src/harness/jsonrpc-stdio.js +7 -0
  35. package/dist/packages/work/src/harness/permissions.d.ts +16 -3
  36. package/dist/packages/work/src/harness/permissions.js +102 -13
  37. package/dist/packages/work/src/harness/session-store.d.ts +27 -9
  38. package/dist/packages/work/src/harness/session-store.js +271 -67
  39. package/dist/packages/work/src/hold/mcp.d.ts +25 -5
  40. package/dist/packages/work/src/hold/mcp.js +126 -11
  41. package/dist/packages/work/src/hub/client.d.ts +2 -1
  42. package/dist/packages/work/src/hub/client.js +14 -1
  43. package/dist/packages/work/src/hub/types.d.ts +69 -33
  44. package/dist/packages/work/src/hub/types.js +8 -10
  45. package/dist/packages/work/src/mcp/server.js +1 -1
  46. package/dist/packages/work/src/owenloop-bin.d.ts +1 -0
  47. package/dist/packages/work/src/owenloop-bin.js +18 -0
  48. package/dist/packages/work/src/roles/agent-run.d.ts +12 -6
  49. package/dist/packages/work/src/roles/agent-run.js +116 -34
  50. package/dist/packages/work/src/roles/exec.d.ts +8 -0
  51. package/dist/packages/work/src/roles/exec.js +27 -2
  52. package/dist/packages/work/src/roles/hold.d.ts +4 -0
  53. package/dist/packages/work/src/roles/hold.js +44 -5
  54. package/dist/packages/work/src/roles/lint.d.ts +3 -4
  55. package/dist/packages/work/src/roles/lint.js +41 -7
  56. package/dist/packages/work/src/roles/settings.js +5 -0
  57. package/dist/packages/work/src/roles/shift.d.ts +1 -0
  58. package/dist/packages/work/src/roles/shift.js +23 -1
  59. package/dist/packages/work/src/settings/settings.d.ts +17 -1
  60. package/dist/packages/work/src/settings/settings.js +39 -0
  61. package/dist/packages/work/src/shift/loop.d.ts +14 -3
  62. package/dist/packages/work/src/shift/loop.js +447 -164
  63. package/dist/packages/work/src/shift/protocol.d.ts +4 -0
  64. package/dist/packages/work/src/shift/runtime.d.ts +2 -0
  65. package/dist/packages/work/src/shift/runtime.js +66 -3
  66. package/dist/packages/work/src/shift/spawn.d.ts +41 -18
  67. package/dist/packages/work/src/shift/spawn.js +89 -31
  68. package/dist/packages/work/src/shift/state.d.ts +53 -51
  69. package/dist/packages/work/src/shift/state.js +325 -102
  70. package/dist/packages/work/src/submit-proof.d.ts +36 -0
  71. package/dist/packages/work/src/submit-proof.js +93 -0
  72. package/dist/src/add.d.ts +64 -259
  73. package/dist/src/add.js +81 -615
  74. package/dist/src/archive.d.ts +97 -0
  75. package/dist/src/archive.js +591 -0
  76. package/dist/src/bundle/index.d.ts +64 -0
  77. package/dist/src/bundle/index.js +508 -0
  78. package/dist/src/bundle/manifest.d.ts +50 -0
  79. package/dist/src/bundle/manifest.js +435 -0
  80. package/dist/src/bundle/runtime.d.ts +26 -0
  81. package/dist/src/bundle/runtime.js +64 -0
  82. package/dist/src/bundle/tar.d.ts +62 -0
  83. package/dist/src/bundle/tar.js +238 -0
  84. package/dist/src/bundle/types.d.ts +105 -0
  85. package/dist/src/bundle/types.js +21 -0
  86. package/dist/src/cli.d.ts +90 -20
  87. package/dist/src/cli.js +2407 -365
  88. package/dist/src/credentials.d.ts +4 -3
  89. package/dist/src/credentials.js +21 -19
  90. package/dist/src/crypto/allowed-signers.d.ts +68 -0
  91. package/dist/src/crypto/allowed-signers.js +307 -0
  92. package/dist/src/crypto/canonical.d.ts +8 -0
  93. package/dist/src/crypto/canonical.js +92 -0
  94. package/dist/src/crypto/chain.d.ts +65 -0
  95. package/dist/src/crypto/chain.js +377 -0
  96. package/dist/src/crypto/dsse.d.ts +209 -0
  97. package/dist/src/crypto/dsse.js +282 -0
  98. package/dist/src/crypto/enrollment.d.ts +78 -0
  99. package/dist/src/crypto/enrollment.js +186 -0
  100. package/dist/src/crypto/index.d.ts +48 -0
  101. package/dist/src/crypto/index.js +44 -0
  102. package/dist/src/crypto/keys.d.ts +258 -0
  103. package/dist/src/crypto/keys.js +908 -0
  104. package/dist/src/crypto/org-root.d.ts +33 -0
  105. package/dist/src/crypto/org-root.js +78 -0
  106. package/dist/src/crypto/origin-rules.d.ts +56 -0
  107. package/dist/src/crypto/origin-rules.js +143 -0
  108. package/dist/src/crypto/policy-floor.d.ts +98 -0
  109. package/dist/src/crypto/policy-floor.js +353 -0
  110. package/dist/src/crypto/records.d.ts +316 -0
  111. package/dist/src/crypto/records.js +157 -0
  112. package/dist/src/crypto/scope.d.ts +49 -0
  113. package/dist/src/crypto/scope.js +83 -0
  114. package/dist/src/crypto/ssh.d.ts +171 -0
  115. package/dist/src/crypto/ssh.js +396 -0
  116. package/dist/src/crypto/submission.d.ts +33 -0
  117. package/dist/src/crypto/submission.js +47 -0
  118. package/dist/src/crypto/trust-roots.d.ts +29 -0
  119. package/dist/src/crypto/trust-roots.js +39 -0
  120. package/dist/src/crypto/verify-consumed.d.ts +67 -0
  121. package/dist/src/crypto/verify-consumed.js +259 -0
  122. package/dist/src/crypto/verify-origin.d.ts +50 -0
  123. package/dist/src/crypto/verify-origin.js +151 -0
  124. package/dist/src/crypto/verify-publication.d.ts +60 -0
  125. package/dist/src/crypto/verify-publication.js +150 -0
  126. package/dist/src/defs.d.ts +24 -1
  127. package/dist/src/defs.js +145 -25
  128. package/dist/src/engine.d.ts +99 -3
  129. package/dist/src/engine.js +238 -34
  130. package/dist/src/factory.d.ts +16 -0
  131. package/dist/src/factory.js +16 -4
  132. package/dist/src/global-config.d.ts +59 -0
  133. package/dist/src/global-config.js +84 -0
  134. package/dist/src/hub.d.ts +75 -21
  135. package/dist/src/hub.js +123 -40
  136. package/dist/src/index.d.ts +17 -2
  137. package/dist/src/index.js +20 -2
  138. package/dist/src/install.d.ts +538 -0
  139. package/dist/src/install.js +1981 -0
  140. package/dist/src/lock.d.ts +15 -57
  141. package/dist/src/lock.js +251 -143
  142. package/dist/src/mcp/serve.d.ts +46 -22
  143. package/dist/src/mcp/serve.js +167 -44
  144. package/dist/src/mcp/server.js +12 -1
  145. package/dist/src/model.d.ts +15 -0
  146. package/dist/src/model.js +111 -1
  147. package/dist/src/order-resolver.d.ts +184 -0
  148. package/dist/src/order-resolver.js +252 -0
  149. package/dist/src/package-version.d.ts +10 -0
  150. package/dist/src/package-version.js +42 -0
  151. package/dist/src/paths.d.ts +14 -1
  152. package/dist/src/paths.js +18 -0
  153. package/dist/src/schemas/enrollment-grant.v1.schema.json +90 -0
  154. package/dist/src/schemas/index.d.ts +1416 -0
  155. package/dist/src/schemas/index.js +31 -0
  156. package/dist/src/schemas/order.v1.schema.json +106 -0
  157. package/dist/src/schemas/origin.v1.schema.json +68 -0
  158. package/dist/src/schemas/policy-floor.v1.schema.json +29 -0
  159. package/dist/src/schemas/publication.v1.schema.json +31 -0
  160. package/dist/src/schemas/revocation.v1.schema.json +39 -0
  161. package/dist/src/schemas/submission.v1.schema.json +61 -0
  162. package/dist/src/store/def-source.d.ts +52 -0
  163. package/dist/src/store/def-source.js +335 -0
  164. package/dist/src/store/index-file.d.ts +69 -0
  165. package/dist/src/store/index-file.js +151 -0
  166. package/dist/src/store/index.d.ts +27 -0
  167. package/dist/src/store/index.js +22 -0
  168. package/dist/src/store/ingestor.d.ts +43 -0
  169. package/dist/src/store/ingestor.js +211 -0
  170. package/dist/src/store/install.d.ts +228 -0
  171. package/dist/src/store/install.js +555 -0
  172. package/dist/src/store/instruction-source.d.ts +49 -0
  173. package/dist/src/store/instruction-source.js +298 -0
  174. package/dist/src/store/origin-evidence.d.ts +21 -0
  175. package/dist/src/store/origin-evidence.js +66 -0
  176. package/dist/src/store/pre-commit-verifier.d.ts +68 -0
  177. package/dist/src/store/pre-commit-verifier.js +418 -0
  178. package/dist/src/store/resolve.d.ts +138 -0
  179. package/dist/src/store/resolve.js +363 -0
  180. package/dist/src/store/types.d.ts +288 -0
  181. package/dist/src/store/types.js +356 -0
  182. package/dist/src/types.d.ts +45 -11
  183. package/dist/src/untar.d.ts +8 -23
  184. package/dist/src/untar.js +10 -136
  185. package/docs/agent-runner.md +86 -0
  186. package/docs/authoring.md +57 -6
  187. package/docs/bundles.md +250 -0
  188. package/docs/cli.md +1171 -97
  189. package/docs/crypto.md +816 -0
  190. package/docs/design.md +342 -20
  191. package/docs/embedding.md +132 -25
  192. package/docs/wire-contracts.md +380 -0
  193. package/package.json +6 -1
  194. package/plugins/claude-code/.claude-plugin/marketplace.json +14 -0
  195. package/plugins/claude-code/plugin/.claude-plugin/plugin.json +8 -0
  196. package/plugins/claude-code/plugin/.mcp.json +14 -0
  197. package/plugins/claude-code/plugin/hooks/hooks.json +26 -0
  198. package/plugins/claude-code/plugin/hooks/session-end.sh +40 -0
  199. package/plugins/claude-code/plugin/hooks/session-start.sh +52 -0
  200. package/plugins/claude-code/plugin/skills/author/SKILL.md +181 -0
  201. package/plugins/claude-code/plugin/skills/conduct/SKILL.md +106 -0
  202. package/plugins/claude-code/plugin/skills/shift/SKILL.md +105 -0
  203. package/plugins/codex/.agents/plugins/marketplace.json +21 -0
  204. package/plugins/codex/plugins/owenloop/.codex-plugin/plugin.json +7 -0
  205. package/plugins/codex/plugins/owenloop/.mcp.json +20 -0
  206. package/plugins/codex/plugins/owenloop/hooks/hooks.json +26 -0
  207. package/plugins/codex/plugins/owenloop/hooks/session-end.sh +40 -0
  208. package/plugins/codex/plugins/owenloop/hooks/session-start.sh +52 -0
  209. package/plugins/codex/plugins/owenloop/skills/author/SKILL.md +181 -0
  210. package/plugins/codex/plugins/owenloop/skills/conduct/SKILL.md +106 -0
  211. package/plugins/codex/plugins/owenloop/skills/shift/SKILL.md +105 -0
@@ -0,0 +1,52 @@
1
+ #!/bin/sh
2
+ # SessionStart hook: report a CLI/plugin version mismatch before MCP calls begin.
3
+ #
4
+ # The root CLI has no --version flag, so ask its MCP initialize response for the
5
+ # server version instead. Read the plugin manifest at runtime so this hook does
6
+ # not add another copied version literal, and always exit 0 so a diagnostic never
7
+ # blocks session startup.
8
+
9
+ plugin_root="${CLAUDE_PLUGIN_ROOT:-}"
10
+ if [ -z "$plugin_root" ]; then
11
+ exit 0
12
+ fi
13
+
14
+ manifest="$plugin_root/.claude-plugin/plugin.json"
15
+ if [ ! -f "$manifest" ]; then
16
+ manifest="$plugin_root/.codex-plugin/plugin.json"
17
+ fi
18
+ if [ ! -f "$manifest" ]; then
19
+ exit 0
20
+ fi
21
+
22
+ # Extract the first flat JSON version field without jq (not guaranteed installed).
23
+ plugin_version="$(tr ',' '\n' < "$manifest" | sed -n 's/[^\"]*"version"[[:space:]]*:[[:space:]]*"\([^\"]*\)".*/\1/p' | sed -n '1p')"
24
+ if [ -z "$plugin_version" ]; then
25
+ echo "owenloop plugin version could not be determined. Run: owenloop setup" >&2
26
+ exit 0
27
+ fi
28
+
29
+ if ! command -v owenloop >/dev/null 2>&1; then
30
+ echo "owenloop CLI was not found on PATH. Run: owenloop setup" >&2
31
+ exit 0
32
+ fi
33
+
34
+ # The initialize response is the root CLI's supported version surface. It is
35
+ # also safe for old CLIs: one that lacks `mcp` fails below and gets a diagnosis.
36
+ initialize_request='{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18"}}'
37
+ # An explicit probe origin bypasses credential-store discovery. The placeholder
38
+ # is scoped to this subprocess, so the probe is deterministic and cannot spend
39
+ # the MCP enrollment probe timeout on a real hub network request.
40
+ probe_hub='https://plugin-version-probe.invalid'
41
+ cli_output="$(printf '%s\n' "$initialize_request" | OWENLOOP_HUB="$probe_hub" owenloop mcp 2>&1)"
42
+ cli_status=$?
43
+ cli_version="$(printf '%s\n' "$cli_output" | tr ',' '\n' | sed -n 's/[^\"]*"version"[[:space:]]*:[[:space:]]*"\([^\"]*\)".*/\1/p' | sed -n '1p')"
44
+ if [ "$cli_status" -ne 0 ] || [ -z "$cli_version" ]; then
45
+ echo "owenloop plugin $plugin_version could not determine the owenloop CLI version. Run: owenloop setup" >&2
46
+ exit 0
47
+ fi
48
+
49
+ if [ "$plugin_version" != "$cli_version" ]; then
50
+ echo "owenloop plugin $plugin_version does not match owenloop CLI $cli_version. Run: owenloop setup" >&2
51
+ fi
52
+ exit 0
@@ -0,0 +1,181 @@
1
+ ---
2
+ name: author
3
+ description: Author a new owenloop workflow through owenloop's create_workflow gate — interview, draft YAML, validate, approve. Use when asked to design/create/author a new workflow for owenloop.
4
+ allowed-tools: mcp__plugin_owenloop_owenloop__create_workflow, mcp__plugin_owenloop_owenloop__start_run, mcp__owenloop__create_workflow, mcp__owenloop__start_run
5
+ ---
6
+
7
+ # author
8
+
9
+ Authoring through the service. You draft the YAML; the engine (via
10
+ `create_workflow`) is the validator, not you. Never trust your own read of
11
+ the grammar over what the engine actually accepts.
12
+
13
+ ## Tool names
14
+
15
+ This plugin attaches the owenloop MCP server, so its tools show up namespaced as
16
+ `mcp__plugin_owenloop_owenloop__<tool>`, e.g.
17
+ `mcp__plugin_owenloop_owenloop__create_workflow` and
18
+ `mcp__plugin_owenloop_owenloop__start_run`. If tool calls by that name fail,
19
+ run `claude mcp list` to confirm the real names before proceeding.
20
+
21
+ ## Steps
22
+
23
+ 1. **Interview the human** for the goal, in plain English. Ask what the
24
+ steps are, what each step needs and produces, and where a human needs to
25
+ sign off. Do not show YAML during this conversation.
26
+ 2. **Draft the def YAML yourself** — name, inputs, steps, consumes/produces,
27
+ schemas on outputs where it makes sense to enforce a shape.
28
+ 3. **Call `create_workflow`** with the draft.
29
+ 4. **If errors come back:** fix the YAML and resubmit. This is a loop — the
30
+ engine's error text is ground truth, not a suggestion. Cap it at ~5
31
+ attempts; if still failing, show the human the raw error and the current
32
+ draft and ask for help.
33
+ 5. **On success:** show the human the returned mermaid diagram and a
34
+ plain-English step list (not the raw YAML) for approval.
35
+ 6. **If any step carries `x.harness`:** ask the human to run `owenloop work lint
36
+ <workflow-name | path>` in their own shell. This skill grants no Bash tool, so
37
+ you cannot run the command yourself. Exit `0` is clean or warnings-only; exit
38
+ `1` means the definition has an error to fix and resubmit through
39
+ `create_workflow`. Runtime permission preflight remains authoritative because
40
+ a later CLI or environment override can select a different adapter.
41
+ 7. **Offer `start_run`** once approved and free of lint errors, with any inputs
42
+ the human already knows they want to seed.
43
+
44
+ ## Step Agent dispatch and `x.harness`
45
+
46
+ `owenloop work agent-run` is the only supported dispatcher for workflow Step
47
+ Agents. The Shift daemon starts an `agent-run` child for each agent order. The
48
+ `conduct` and `shift` skills supervise the Shift daemon; plugin-packaged workers
49
+ do not dispatch workflow steps.
50
+
51
+ A step carries adapter selection and policy under the fixed, vendor-neutral
52
+ `x.harness` map:
53
+
54
+ ```yaml
55
+ x:
56
+ harness:
57
+ id: claude-code
58
+ tools: [Read, Glob]
59
+ disallowedTools: [Bash]
60
+ filesystem: read-only
61
+ network: owenloop-only
62
+ permissionMode: default
63
+ maxTurns: 20
64
+ model: sonnet
65
+ effort: high
66
+ ```
67
+
68
+ `id` is optional. When `id` is absent, the runtime may select the registered
69
+ default adapter; `--harness` and `OWENLOOP_HARNESS` overrides take precedence
70
+ over the definition. Keep policy in `x.harness` even when `id` is absent.
71
+
72
+ When a producer declares native `judges:`, put `x.harness` on the producer
73
+ step, not on each judge entry. Every synthesized judge automatically inherits
74
+ an independent deep clone of the producer's complete `x` map. An absent
75
+ producer `x` stays absent. A judge entry's first-class `model` still wins over
76
+ an inherited `x.harness.model`. CLI and environment harness overrides still
77
+ select the final adapter, and that final adapter must pass policy preflight
78
+ before either a judge cold start or a judge resume.
79
+
80
+ The eight neutral fields are:
81
+
82
+ - `tools`: a comma-separated string or string array. An absent field declares no
83
+ allow-list. `tools: []` explicitly disables all built-in tools on an adapter
84
+ that supports tool allow-lists.
85
+ - `disallowedTools`: a comma-separated string or string array. `tools` and
86
+ `disallowedTools` must not overlap.
87
+ - `filesystem`: `read-only`, `workspace-write`, or `unrestricted`.
88
+ - `network`: `owenloop-only` or `unrestricted`.
89
+ - `permissionMode`: a non-empty adapter-native mode listed below.
90
+ - `maxTurns`: a positive integer.
91
+ - `model`: a non-empty model id. A first-class step `model` field takes
92
+ precedence when present.
93
+ - `effort`: a non-empty effort value accepted by the selected adapter.
94
+
95
+ Do not put generated `name` or `description` fields in `x.harness`. Unknown
96
+ keys remain opaque adapter extension data; an adapter may interpret or warn
97
+ about those keys, but unknown keys do not create a neutral security boundary.
98
+
99
+ ### Exact adapter support
100
+
101
+ | Policy | Claude Code adapter | Codex adapter |
102
+ |---|---|---|
103
+ | absent `tools` | Preserves the normal built-in-tool default unless `filesystem` or `network` requires an audited set. | Supported only because no tool list was authored. |
104
+ | `tools: []` or a non-empty `tools` list | Enforced through both the available and auto-allowed tool sets. Settings, skills, and external MCP are isolated so unlisted tools cannot widen the surface. Authored external MCP tool names are refused. | Refused because Codex cannot enforce a per-thread built-in-tool list. |
105
+ | `disallowedTools` | Enforced. A deny that blocks the born-bound Owenloop `get_order` or `submit` control tools is refused. | Refused. |
106
+ | `filesystem: read-only` | Enforced. With unrestricted network the audited defaults are `Read`, `Glob`, `Grep`, `WebFetch`, and `WebSearch`; with `network: owenloop-only` the defaults narrow to `Read`, `Glob`, and `Grep`. | Refused because app-server configuration layers sit outside the thread sandbox. |
107
+ | `filesystem: workspace-write` | Refused because the adapter cannot prove an exact workspace boundary. | Refused because app-server configuration layers sit outside the thread sandbox. |
108
+ | `filesystem: unrestricted` | Supported. | Supported as sandbox `danger-full-access`; a conflicting vendor `sandbox` extension is refused. |
109
+ | `network: owenloop-only` | Enforced by isolating settings, skills, MCP, and built-in tools while retaining only the born-bound Owenloop control plane. | Refused. |
110
+ | `network: unrestricted` | Supported. | Supported with `filesystem: unrestricted`, or with omitted `filesystem` and sandbox `workspace-write` or `danger-full-access`; refused with sandbox `read-only`. |
111
+ | `permissionMode` | `default`, `acceptEdits`, `bypassPermissions`, `plan`, `dontAsk`, or `auto`. | `untrusted`, `on-request`, or `never`. |
112
+ | `maxTurns` | Enforced. | Refused because Codex app-server has no thread or turn limit parameter. |
113
+ | `model` | Supported. | Supported. |
114
+ | `effort` | `low`, `medium`, `high`, `xhigh`, or `max`. | Passed to Codex `turn/start`; use a value supported by the selected Codex runtime. |
115
+
116
+ The Claude Code adapter also refuses a `read-only` allow-list containing
117
+ non-read-only tools, an `owenloop-only` allow-list containing network-capable or
118
+ unaudited tools, and any policy that denies the required Owenloop control tools.
119
+ For every restricted Claude Code cold start or resume, the mounted MCP child
120
+ positively registers exactly `get_order` and `submit`. `allowedTools` only
121
+ auto-allows calls; `allowedTools` does not remove registrations from MCP
122
+ `tools/list`. The adapter also denies `mcp__owenloop__reject` as defense in depth.
123
+ Outside Claude Code isolation, the default work-holder MCP child registers
124
+ `get_order`, `submit`, and `reject`.
125
+
126
+ Every defined Codex `maxTurns` value is refused before cold start or resume. The
127
+ Codex adapter starts no app-server process, and `agent-run` releases the held
128
+ claim and exits non-zero.
129
+
130
+ ### Fail-closed rule
131
+
132
+ `x.harness` security restrictions are enforced policy. Runtime preflight must
133
+ prove that the final selected adapter can implement every authored restriction
134
+ exactly. An unsupported adapter restriction is refused: no model SDK, CLI, or
135
+ app-server process starts, the worker releases the held claim, and `agent-run`
136
+ exits non-zero with every refusal reason.
137
+
138
+ `advisory.tools` is model guidance only. `advisory.tools` does not configure an
139
+ adapter, does not remove a provider tool, and is not a security boundary.
140
+
141
+ ## Rules
142
+
143
+ - Never show raw YAML unless the human explicitly asks to see it.
144
+ - The engine's error text is ground truth — pass it through verbatim, don't
145
+ paraphrase or guess at a fix without reading it.
146
+ - Schemas on outputs are encouraged, not optional busywork — that's the
147
+ actual enforcement mechanism (tight schemas catch sloppy workers; loose
148
+ ones are a deliberate choice for a happy-path step, not an oversight).
149
+ - **Never write a Crew name in a def.** A def is portable; a Crew name is one
150
+ org's deployment fact. `personal:` names in particular are rejected as
151
+ Capability names.
152
+ - **Default to no `capabilities:`.** A step with no `capabilities:` is
153
+ def-silent and routes to the run's default Crew — the starter's personal Crew
154
+ unless the starter chooses otherwise. That is the normal case; most defs
155
+ should carry no Capability requirement.
156
+ - **Use a Capability only when a step genuinely needs a specific fleet**
157
+ (`gpu`, `repo-access` — a routing requirement, not a team name, machine, or
158
+ Crew). An admin binds that Capability to one or more Crews per hub; the def
159
+ never learns which Crews.
160
+ - **An unbound Capability fails the whole run at start**, not at the step:
161
+ `start_run` is refused with `capability_route_invalid` naming the Capability
162
+ and the fix. If you introduce a Capability, tell the human that the hub needs
163
+ `owenloop capability bind <capability> <crew>` (or Console → Settings →
164
+ Capabilities) before the def can run.
165
+
166
+ ## Setup (once per machine)
167
+
168
+ - Install the plugin (adds the owenloop MCP server automatically). Until this
169
+ pack is published to a public marketplace, add it via its local path in
170
+ this repo:
171
+ ```
172
+ claude plugin marketplace add ./packages/driver-claude-code/marketplace
173
+ claude plugin install owenloop@owenloop
174
+ ```
175
+ - Run `owenloop setup` once — its single loopback OAuth is the only browser
176
+ step; target a non-prod owenloop server with `owenloop setup --hub <origin>`.
177
+ The owenloop stdio server then spawns pre-trusted in every session — no
178
+ per-session consent, no token to copy for interactive use.
179
+ Headless/CI setups instead add a separate server with an admin-minted
180
+ `olp_...` token (`OWENLOOP_TOKEN`) — see the pack's `README.md` for the
181
+ full install + connection walkthrough, both paths.
@@ -0,0 +1,106 @@
1
+ ---
2
+ name: conduct
3
+ description: Drive one owenloop workflow with a scoped shift, a blocking CLI park, status checks, and human gate relay. Use when asked to run, conduct, or drive an owenloop workflow.
4
+ allowed-tools: Bash(owenloop:*)
5
+ ---
6
+
7
+ # conduct
8
+
9
+ You are a **one-workflow supervisor**. You do no workflow step work, hold no
10
+ order, claim no order, dispatch no order, submit no step output, repair no run,
11
+ and never close a run. You prepare and identify exactly one target workflow,
12
+ keep a compatible shift attended, relay human gates, and stop conducting only
13
+ when the target workflow is closed.
14
+
15
+ ## Required safety clauses
16
+
17
+ These clauses control the conduct loop. Do not replace them with a timer, a
18
+ summary, or a claim that the workflow is complete.
19
+
20
+ 1. **Starting is not running.** Starting a daemon, announcing a shift, or
21
+ completing one poll does not mean the shift is being run; stopping after
22
+ startup abandons the shift.
23
+ 2. **Ban the substitutes.** Never schedule a wakeup, reminder, cron job,
24
+ scheduler, timeout, interval, or cadence; never say "I'll check back later";
25
+ never end the turn to wait. `--wait 90` is the blocking CLI park itself, not
26
+ an external timer or scheduler.
27
+ 3. **The blocking call is the wait.** `owenloop shift next --wait 90` blocks
28
+ until there is something to report or its bounded park returns. Blocking is
29
+ standing by. When the command returns, process the result, run
30
+ `owenloop status <wf>` for the captured target, handle that status, and only
31
+ then run the command again if the target remains open. A normal or empty
32
+ event result never skips the required status check.
33
+ 4. **Judging rule.** For a goal shaped like "run a shift on <crew>", the goal
34
+ is unmet until a human explicitly asks to end that shift and `owenloop shift
35
+ end` confirms the end. Poll count, elapsed time, daemon startup, and idle
36
+ returns never satisfy the shift goal. For conduct, the conduct request is
37
+ complete only when `owenloop status <wf>` reports the target workflow closed.
38
+
39
+ A no-work or empty event return is normal duty, not a blocking condition, not
40
+ an error, and never a reason for Codex to mark the goal `blocked`.
41
+
42
+ ## Prepare exactly one target
43
+
44
+ 1. Run `owenloop work prepare <workflow>`. If the command exits non-zero,
45
+ relay its output verbatim and stop. Do not invent a workaround or continue
46
+ against an unprepared definition.
47
+ 2. Obtain an explicit crew scope before starting a new daemon. Run
48
+ `owenloop shift status` and inspect the daemon's exact served scope.
49
+ Reuse a compatible existing daemon when possible.
50
+ - If the status reports no daemon, start `owenloop shift start <crew...>` as
51
+ a background process owned by the current session and state that the
52
+ daemon ends with the session. A human-started terminal or tmux daemon is
53
+ the durable alternative.
54
+ - If the status reports an existing daemon serving an incompatible scope,
55
+ relay the exact status and ask the human how to proceed before ending,
56
+ restarting, or changing that daemon's scope. Do not start a second daemon
57
+ or silently repurpose the existing daemon. Changing scope affects the
58
+ whole shift.
59
+ Never silently use all crews; use `owenloop shift start --all` only after
60
+ explicit confirmation.
61
+ 3. Create the target with `owenloop create <workflow>` unless the human
62
+ supplied an existing `wf_...` instance. Capture exactly one target workflow
63
+ id. Treat Capability-route and start errors as verbatim human-facing failures;
64
+ do not run admin repair commands.
65
+
66
+ ## Attend the target through one blocking park
67
+
68
+ Run one park at a time: keep only one blocking `owenloop shift next --wait 90`
69
+ call outstanding. After each park returns, follow this order before any next
70
+ park:
71
+
72
+ 1. Process and report the returned capacity and events. Relay command failures
73
+ verbatim. Cap every status or log excerpt at the last **4 lines** per item;
74
+ never dump full logs into chat.
75
+ 2. Run `owenloop status <wf>` for the captured target, including after a normal
76
+ or empty event result.
77
+ 3. Handle the target status.
78
+ - If the status reports `waitingOnCapabilities` as `[{ step, capabilities }]`,
79
+ name the exact step and exact Capability or Capabilities to the human.
80
+ Explain that the run resumes when an admin binds the Capability to a Crew.
81
+ Do not restart the run, repair the route, or treat the run as a crashed
82
+ Worker.
83
+ <!-- Test-fixture compatibility only: `packages/driver-claude-code/test/plugin-shape.test.ts` still searches this file for the retired literals `waitingOnLabels` as `[{ step, labels }]`, `name the exact step and exact label or labels`, `run resumes when an admin re-binds the label`, and `Do not restart the run, repair the binding, or treat the run as a crashed worker.` The live field and guidance above are authoritative: `waitingOnCapabilities`, Capabilities, Crews, and Worker. This comment is not product guidance. -->
84
+ - Relay pending gates through `owenloop provide <wf> <name> --value '<json>'`
85
+ only after the human answers. Pass the JSON object shape required by the
86
+ CLI. Never infer, synthesize, or auto-answer a gate.
87
+ - If the status reports the captured target workflow closed, stop conducting.
88
+ - If the target remains open, only then run the next blocking park.
89
+
90
+ - If a second park is attempted, relay the CLI error verbatim and wait for the
91
+ outstanding call; never issue another park.
92
+ - `running: 0`, an empty events list, and one idle park are not completion.
93
+ Completion is only `owenloop status <wf>` reporting that the captured target
94
+ workflow is closed.
95
+
96
+ ## Daemon cleanup and boundaries
97
+
98
+ Do not end a pre-existing shared shift. If conduct started a session-owned
99
+ daemon solely for this target, state that lifecycle at startup and run
100
+ `owenloop shift end` after the target closes. Ending that session-owned daemon
101
+ is cleanup; it does not change the separate rule that a user-requested standing
102
+ shift requires an explicit human end request.
103
+
104
+ Never claim, dispatch, submit, repair, or close a run. The supervisor only
105
+ prepares and identifies the target, keeps the scoped shift attended, checks the
106
+ target status, and relays human decisions.
@@ -0,0 +1,105 @@
1
+ ---
2
+ name: shift
3
+ description: Go on a standing owenloop shift for one or more crews, keep the blocking CLI park running, and relay human gates. Use when asked to go on shift, start a shift, or attend a crew (`/shift start <crew>`).
4
+ allowed-tools: Bash(owenloop:*)
5
+ ---
6
+
7
+ # shift
8
+
9
+ You are a **standing shift supervisor**. You do no workflow step work, hold no
10
+ order, and never submit or close a run. You keep one owenloop shift on duty
11
+ for the human-named crews and relay events and human gates. This is a standing
12
+ posture: the shift continues until the human explicitly ends it. It is not a
13
+ one-workflow conduct request.
14
+
15
+ ## Goal wiring
16
+
17
+ Determine the harness goal wiring before starting the shift.
18
+
19
+ - **Claude Code:** tell the human to type exactly `/goal run a shift on <crew>`.
20
+ The model cannot invoke `/goal` itself because only the human can type that
21
+ built-in command. Do not claim that this skill installed the goal.
22
+ - **Codex:** invoke `create_goal` with objective exactly `the shift on <crew> has been explicitly stopped by the human`, then enter the same
23
+ Bash loop. `allowed-tools` is Claude Code frontmatter; Codex follows this
24
+ Bash-only body.
25
+
26
+ ## Required safety clauses
27
+
28
+ These clauses control the shift. Do not replace them with a timer, a summary,
29
+ or a claim that the shift is complete.
30
+
31
+ 1. **Starting is not running.** Starting a daemon, announcing a shift, or
32
+ completing one poll does not mean the shift is being run; stopping after
33
+ startup abandons the shift.
34
+ 2. **Ban the substitutes.** Never schedule a wakeup, reminder, cron job,
35
+ scheduler, timeout, interval, or cadence; never say "I'll check back later";
36
+ never end the turn to wait. `--wait 90` is the blocking CLI park itself, not
37
+ an external timer or scheduler.
38
+ 3. **The blocking call is the wait.** `owenloop shift next --wait 90` blocks
39
+ until there is something to report or its bounded park returns. Blocking is
40
+ standing by. When the command returns, process the result and immediately
41
+ run the command again unless the explicit completion condition has occurred.
42
+ 4. **Judging rule.** For a goal shaped like "run a shift on <crew>", the goal
43
+ is unmet until a human explicitly asks to end that shift and `owenloop shift
44
+ end` confirms the end. Poll count, elapsed time, daemon startup, and idle
45
+ returns never satisfy the shift goal.
46
+
47
+ A no-work or empty event return is normal duty, not a blocking condition, not
48
+ an error, and never a reason for Codex to mark the goal `blocked`.
49
+
50
+ ## Start the explicitly scoped shift
51
+
52
+ 1. Require at least one named crew. If the human names no crew, ask whether
53
+ the human means all crews. Use `owenloop shift start --all` only after the
54
+ human explicitly confirms all crews. Never encode all crews as an empty
55
+ argument or a silent default.
56
+ 2. Run `owenloop shift status` and inspect the daemon's exact served scope.
57
+ Reuse a compatible existing daemon when possible.
58
+ - If the status reports no daemon, start `owenloop shift start <crew...>` as
59
+ a background process owned by the current session and state that the
60
+ daemon ends with the session. A human-started terminal or tmux daemon is
61
+ the durable alternative.
62
+ - If the status reports an existing daemon serving an incompatible scope,
63
+ relay the exact status and ask the human how to proceed before ending,
64
+ restarting, or changing that daemon's scope. Do not start a second daemon
65
+ or silently repurpose the existing daemon. Changing scope affects the
66
+ whole shift.
67
+
68
+ ## Park one call at a time
69
+
70
+ Run exactly one `owenloop shift next --wait 90` call at a time. The command is
71
+ the blocking park. When the command returns, parse and report its capacity and
72
+ events, then immediately run it again after normal or empty results. There is
73
+ one park at a time; never overlap parks.
74
+
75
+ - Never issue a second park while the first one is outstanding. If the CLI
76
+ rejects an overlapping park, relay the error verbatim and wait for the
77
+ outstanding call; never issue another park.
78
+ - Relay command failures verbatim. Do not paraphrase, hide, or repair a failed
79
+ command.
80
+ - Cap any status or log excerpt at the last **4 lines** per item. Do not dump
81
+ full logs into chat.
82
+ - A daemon process exit or a `shift next` no-daemon error may be relayed and
83
+ restarted only within the human-approved crew scope.
84
+
85
+ ## Relay gates and events
86
+
87
+ When a poll reports an event, tell the human what happened. For a gate event,
88
+ present the gate question to the human and wait for the human's answer. Run
89
+ `owenloop provide <wf> <name> --value '<json>'` only with the human's answer,
90
+ using the JSON object shape required by the CLI. Never synthesize, infer, or
91
+ auto-answer a gate.
92
+
93
+ The shift does not discover arbitrary gates before the gate-event contract is
94
+ available. Do not claim that a daemon-local event is a hub gate.
95
+
96
+ ## End condition
97
+
98
+ Run `owenloop shift end` only after the human explicitly asks to end the shift.
99
+ A confirmed ended event or status is the shift completion condition. Until that
100
+ confirmation, keep parking; an idle return, zero running workers, elapsed time,
101
+ or daemon startup is not completion.
102
+
103
+ Do not start a workflow, claim an order, dispatch an order, submit step output,
104
+ repair a run, or close a run. The shift supervisor only keeps the scoped shift
105
+ attended and relays information and human decisions.