@vtxmacro/cli 2026.8.10 → 2026.8.11

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 (3) hide show
  1. package/README.md +52 -3
  2. package/bin/vtx.js +5311 -4531
  3. package/package.json +1 -1
package/README.md CHANGED
@@ -39,7 +39,13 @@ vtx-mcp
39
39
 
40
40
  The CLI and MCP server use the same saved token and environment variables.
41
41
 
42
- ## Subscription inference host
42
+ ## Agent inference hosts
43
+
44
+ Both paths use a separate least-privilege `insights:inference` OAuth grant.
45
+ VTX never receives an agent vendor password, OAuth token, or subscription
46
+ credential.
47
+
48
+ ### Automated Codex host
43
49
 
44
50
  ```bash
45
51
  vtx inference-host login
@@ -57,13 +63,56 @@ OAuth grant. It also keeps its ChatGPT subscription login in a dedicated private
57
63
  Codex home. Neither credential is copied from or widens the ordinary CLI or Codex
58
64
  session. VTX logout/revoke and Codex logout are intentionally separate operations.
59
65
 
66
+ ### Any compatible agent harness
67
+
68
+ Codex, Claude Code, Cursor, GitHub Copilot, Antigravity, and other agents that can
69
+ run shell commands can use their existing first-party session without a
70
+ VTX-owned vendor adapter:
71
+
72
+ ```bash
73
+ vtx inference-host login
74
+ vtx inference-host agent-connect \
75
+ --adapter <lowercase-harness-id> \
76
+ --model <exact-model-id> \
77
+ --model-label "<display label>" \
78
+ --effort <reasoning-effort> \
79
+ --json
80
+ vtx inference-host agent-next --wait-seconds 50 --json
81
+ ```
82
+
83
+ `agent-next` returns separate exact `system_prompt`, `user_prompt`,
84
+ `context_json`, and `output_schema_json` fields. The agent reasons in its own
85
+ session, then writes one JSON object to `agent-complete` stdin. At minimum that
86
+ object contains a `result` string. If the harness exposes exact token usage,
87
+ include the published usage object; otherwise VTX records usage as explicitly
88
+ unavailable. For `agent-fail`, choose the truthful dispatch state and set
89
+ `retryable` true only when a fresh attempt is safe:
90
+
91
+ ```json
92
+ {"dispatch_outcome":"not_dispatched","failure_category":"adapter","failure_code":"harness_unavailable","retryable":true}
93
+ {"dispatch_outcome":"confirmed_dispatched","failure_category":"adapter","failure_code":"no_reliable_result","retryable":false}
94
+ {"dispatch_outcome":"outcome_unknown","failure_category":"transport","failure_code":"dispatch_confirmation_lost","retryable":false}
95
+ ```
96
+
97
+ Use `not_dispatched` only when no harness inference began,
98
+ `confirmed_dispatched` when it began but returned no trustworthy result, and
99
+ `outcome_unknown` when dispatch itself cannot be confirmed. Pipe exactly one
100
+ object to `vtx inference-host agent-fail --json`.
101
+
102
+ Keep calling `agent-next` while the host should remain available. This same loop
103
+ supports Main, Review, and Screener; Provider response and Decision candidate;
104
+ and normally running VTX Client Mode or Server Mode bots. Server Mode keeps the
105
+ normal VTX call fee. Client Mode has no VTX platform fee. Neither mode falls back
106
+ to another model when the agent host is unavailable.
107
+
60
108
  ## Remove
61
109
 
62
- First revoke the VTX inference grant and remove the dedicated Codex login, then
63
- uninstall the package:
110
+ First revoke the VTX inference grant, then uninstall the package. Automated
111
+ Codex users should also remove the dedicated Codex login:
64
112
 
65
113
  ```bash
66
114
  vtx inference-host revoke
115
+ # Automated Codex host only:
67
116
  vtx inference-host codex-logout
68
117
  npm uninstall -g @vtxmacro/cli
69
118
  ```