@rnbsolucoes/axion-code 0.1.27 → 0.1.29
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
CHANGED
|
@@ -42,6 +42,8 @@ axion-code provider profile set or-free
|
|
|
42
42
|
axion-code provider profile delete or-free
|
|
43
43
|
axion-code provider profile active
|
|
44
44
|
axion-code provider model list openrouter --json
|
|
45
|
+
axion-code provider cost set openrouter --model openai/gpt-5-mini --max-usd 0.25 --input-usd-per-1m 0.25 --output-usd-per-1m 2 --max-output-tokens 4000
|
|
46
|
+
axion-code provider cost status --profile openrouter --model openai/gpt-5-mini --json
|
|
45
47
|
axion-code provider test openrouter google/gemini-2.5-flash-lite
|
|
46
48
|
axion-code permission list --json
|
|
47
49
|
axion-code permission set full_permission
|
|
@@ -61,7 +63,7 @@ axion-code mcp import .\mcp-servers.json
|
|
|
61
63
|
axion-code plugins list --json
|
|
62
64
|
axion-code agents list --json
|
|
63
65
|
axion-code agents run bug-hunter "review this login flow"
|
|
64
|
-
axion-code agents run-many bug-hunter,performance-reviewer "review this login flow" --budget-tokens 2000
|
|
66
|
+
axion-code agents run-many bug-hunter,performance-reviewer "review this login flow" --budget-tokens 2000 --stream
|
|
65
67
|
axion-code learn --json
|
|
66
68
|
axion-code package inspect <path>
|
|
67
69
|
```
|
|
@@ -173,10 +175,21 @@ Axion Code selection is isolated:
|
|
|
173
175
|
%USERPROFILE%\.axion\mcp-servers.json
|
|
174
176
|
%USERPROFILE%\.axion\plugins.json
|
|
175
177
|
%USERPROFILE%\.axion\sub-agents.json
|
|
178
|
+
%USERPROFILE%\.axion\harness\cost-policy.json
|
|
176
179
|
```
|
|
177
180
|
|
|
178
181
|
OpenRouter uses `OPENROUTER_API_KEY` from the environment or a `secret:<NAME>` reference stored in `C:\Israel\Pesoal\secrets.env` (override for tests: `AXION_SECRETS_ENV`). Provider profile config stores only references, never the key value.
|
|
179
182
|
|
|
183
|
+
Provider-specific cost ceilings are stored outside the Desktop profile schema in
|
|
184
|
+
`%USERPROFILE%\.axion\harness\cost-policy.json`. Use `provider cost set` to
|
|
185
|
+
bind a ceiling and token prices to a profile or profile/model pair. The runtime
|
|
186
|
+
blocks a request before provider execution when the estimated input or configured
|
|
187
|
+
maximum output cost would exceed the ceiling. After a response, actual
|
|
188
|
+
input/output usage is converted to cost and surfaced in provider/subagent stream
|
|
189
|
+
telemetry; if actual cost exceeds the ceiling, the turn returns an explicit cost
|
|
190
|
+
ceiling error. A ceiling without token prices is not enforced, because Axion Code
|
|
191
|
+
does not infer provider pricing silently.
|
|
192
|
+
|
|
180
193
|
For test/dev isolation, set `AXION_HOME` to a temporary folder. In normal use,
|
|
181
194
|
leave it unset so Desktop and CLI share the same provider catalog while keeping
|
|
182
195
|
their active provider/model selections independent.
|
|
@@ -239,7 +252,7 @@ unchanged.
|
|
|
239
252
|
|
|
240
253
|
`/mcp` opens an inline MCP list. Enabled servers can be disabled, disabled servers can be enabled, and non-native servers can be uninstalled while preserving valid `mcp-servers.json` formatting.
|
|
241
254
|
|
|
242
|
-
`/agents` lists the native isolated subagents. `/agents run <id> <prompt>` calls one with an isolated prompt while inheriting the active provider/model from the main agent. `/agents run-many <id,id,...> <prompt> [--budget-tokens <n>]` runs up to eight isolated subagents in parallel, then returns their individual outputs plus a compact comparison and aggregate token usage. If the estimated isolated prompt input already exceeds the budget, no provider calls are executed.
|
|
255
|
+
`/agents` lists the native isolated subagents. `/agents run <id> <prompt>` calls one with an isolated prompt while inheriting the active provider/model from the main agent. `/agents run-many <id,id,...> <prompt> [--budget-tokens <n>]` runs up to eight isolated subagents in parallel, streams lifecycle/telemetry events into the TUI while the team runs, then returns their individual outputs plus a compact comparison and aggregate token usage. If the estimated isolated prompt input already exceeds the budget, no provider calls are executed.
|
|
243
256
|
|
|
244
257
|
## MCP, Plugins And Subagents
|
|
245
258
|
|
|
@@ -381,4 +394,4 @@ This is a functional direction MVP, not the full harness:
|
|
|
381
394
|
`tool_result` continuation turn;
|
|
382
395
|
- remote MCP transports, browser, process and unknown tools remain blocked;
|
|
383
396
|
- no executable Pi RPC bridge yet;
|
|
384
|
-
- native subagent execution is prompt-isolated and provider/model-inherited; `agents run-many` provides parallel fan-out, compact comparison, aggregate usage
|
|
397
|
+
- native subagent execution is prompt-isolated and provider/model-inherited; `agents run-many` provides parallel fan-out, compact comparison, aggregate usage, a token budget ceiling and streamed lifecycle/telemetry events. Provider/model cost ceilings are enforced by the shared Axion cost policy.
|
|
Binary file
|
|
Binary file
|