nexarch 0.12.33 → 0.12.35
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/lib/skills.js +35 -4
- package/package.json +1 -1
package/dist/lib/skills.js
CHANGED
|
@@ -104,6 +104,28 @@ skill is the check-in → claim → complete loop for working that queue.
|
|
|
104
104
|
- \`nexarch_check_in\` previews pending commands and draft/proposed
|
|
105
105
|
applications for this agent. It does **not** claim or change anything —
|
|
106
106
|
safe to call any time, including proactively at the start of a session.
|
|
107
|
+
- If you have a **confirmed** application ref — returned by an earlier tool
|
|
108
|
+
call in this session (\`target_entity_key\` from a prior check-in,
|
|
109
|
+
\`nexarch_list_entities\`, \`nexarch_resolve_reference\`), or the human gave
|
|
110
|
+
you the exact ref — pass it as
|
|
111
|
+
\`applicationContext: { accessibleApplicationRefs: ["application:<ref>"] }\`
|
|
112
|
+
on this **first** call to skip a redundant round trip.
|
|
113
|
+
- **Never guess the ref from a project/repo/directory name.** A repository
|
|
114
|
+
named \`veri-viva-platform\` does not mean its application ref is
|
|
115
|
+
\`application:veri_viva_platform\` — the real one might be
|
|
116
|
+
\`application:veri_viva_platform_website\` or anything else; projects and
|
|
117
|
+
applications are different entities with independently-chosen refs (see
|
|
118
|
+
the architecture-graph skill). A wrong-but-plausible-looking ref doesn't
|
|
119
|
+
just fail to claim — it silently filters real pending commands out of the
|
|
120
|
+
response, and check-in will wrongly report nothing pending. If you're not
|
|
121
|
+
certain a ref is correct, call check-in bare instead.
|
|
122
|
+
- Calling bare (no \`applicationContext\`) is always safe and returns every
|
|
123
|
+
pending command's real \`target_entity_key\` — use that to make an
|
|
124
|
+
informed scoped call afterward.
|
|
125
|
+
- If a scoped call's response includes a
|
|
126
|
+
\`pending_commands_outside_requested_scope\` warning, the ref you supplied
|
|
127
|
+
didn't match everything — it's very likely wrong. Re-call bare rather than
|
|
128
|
+
trusting the (incomplete) result.
|
|
107
129
|
- Report exactly what check-in found. Don't reinterpret "check in" as
|
|
108
130
|
registration (\`init-agent\`) or as a general health check
|
|
109
131
|
(\`nexarch_get_governance_summary\`) — they're different actions.
|
|
@@ -112,13 +134,22 @@ skill is the check-in → claim → complete loop for working that queue.
|
|
|
112
134
|
|
|
113
135
|
## Working a command
|
|
114
136
|
|
|
137
|
+
Claiming is a commitment, not a checkbox. Don't claim a command unless
|
|
138
|
+
you're about to do the three steps below in the same turn — a claimed
|
|
139
|
+
command that's never executed or closed out sits there looking done while
|
|
140
|
+
blocking that queue slot, which is worse than never having claimed it.
|
|
141
|
+
|
|
115
142
|
1. Only claim a specific command with \`nexarch_claim_command_by_id\` when the
|
|
116
143
|
human explicitly wants it worked — check-in surfacing a command is not
|
|
117
|
-
itself permission to claim it
|
|
118
|
-
|
|
144
|
+
itself permission to claim it, and neither is check-in itself: claiming
|
|
145
|
+
is a separate, deliberate call you make because you're about to act.
|
|
146
|
+
2. **Execute \`command.resolved_playbook_text\` from the claim response** —
|
|
147
|
+
that field is the actual task. \`command.instructions\` is a different
|
|
148
|
+
field and is almost always \`null\`; don't mistake its emptiness for "no
|
|
149
|
+
work to do."
|
|
119
150
|
3. Close it out: \`nexarch_complete_command\` on success, or
|
|
120
|
-
\`nexarch_fail_command\` with a reason if it couldn't be done. Never
|
|
121
|
-
|
|
151
|
+
\`nexarch_fail_command\` with a reason if it couldn't be done. Never end a
|
|
152
|
+
turn with a command still claimed and nothing else called.
|
|
122
153
|
|
|
123
154
|
## If check-in shows a command you can't claim
|
|
124
155
|
|