openai/codex
1
0
Fork 0
mirror of https://github.com/openai/codex.git synced 2026-10-18 17:18:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

Refresh bundled model metadata and instructions (#47397)

Browse Source 
## What changed

- Switch the GPT-6, GPT-5.6, GPT-5.5, and auto-review catalog entries to `shell_command` and expand their plan availability.
- Mark `gpt-6-astra` as supporting reasoning-effort updates and disable its experimental-context flag. Add explicit capability and instruction defaults to other entries.
- Update Astra's messaging guidance to recognize authorization from explicitly invoked skills or plugins, with attribution in the final response, and simplify its Default-mode clarification instructions.
- Revise auto-review guidance on autonomy, assumptions, and actions within the user's requested scope.
- Refresh seven scenario snapshots to reflect the updated prompts.

GitOrigin-RevId: b7ed8c413f6451008c982024506009b170e6311e
This commit is contained in:
andrewgu-oai
2026-09-23 00:07:56 +00:00
committed by copyberry
parent 2c3295306f
commit 24462234b2
8 changed files with 104 additions and 64 deletions
Download Patch File Download Diff File Expand all files Collapse all files
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__astra_code_mode_call_timing.snap
+2 -5
View File
@@ -30,7 +30,7 @@ Scenario: Astra receives host and handler timings on completed and failed code-m
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -54,15 +54,12 @@ Scenario: Astra receives host and handler timings on completed and failed code-m
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__astra_disabled_executor_skills.snap
+2 -5
View File
@@ -40,7 +40,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -77,15 +77,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
[03] <skills_instructions>
## Skills
A skill is a set of instructions provided through a `SKILL.md` source. Below is the list of skills that ...locators can be expanded using the skill roots table. [hash=f00d6a4d3e192f06]
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__astra_kickoff_remote_compaction_windows.snap
+4 -10
View File
@@ -40,7 +40,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -89,15 +89,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
@@ -214,7 +211,7 @@ Settings: same as window 1
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -268,15 +265,12 @@ Settings: same as window 1
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
06:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__astra_plugin_refresh.snap
+4 -10
View File
@@ -40,7 +40,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -64,15 +64,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
@@ -137,7 +134,7 @@ Settings: same as window 1
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -161,15 +158,12 @@ Settings: same as window 1
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__astra_settings_release_check_tool_shapes.snap
+2 -5
View File
@@ -40,7 +40,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -64,15 +64,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__code_mode_catalog_messages.snap
+2 -5
View File
@@ -15,7 +15,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -63,15 +63,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
01:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/core/tests/suite/snapshots/all__suite__scenarios__multi_agent_catalog_parameters.snap
+2 -5
View File
@@ -40,7 +40,7 @@ Settings:
User authorization and preferences persist across turns. Do not request permission again when the user h... any guidelines provided in skills or external files. [hash=911f91c92c72882c]
<OMITTED 155 LINES; ~4895 TOKENS; hash=fd5ef18a80f62502>
<OMITTED 155 LINES; ~4935 TOKENS; hash=5573869716192298>
- Skill naming: If a plugin contributes skills, those skill entries are prefixed with plugin_name: in the Skills list.
- MCP naming: Plugin-provided MCP tools keep standard MCP identifiers such as mcp__server__tool; use tool provenance to tell which plugin they come from.
@@ -64,15 +64,12 @@ Settings:
Use the `request_user_input` tool only when it is listed in the available tools for this turn.
In Default mode, strongly prefer making reasonable assumptions and executing the user's request rather than stopping to ask questions.
Use the `request_user_input` tool only for optional questions where the answer would materially improve the quality of the work.
If `request_user_input` returns no answers, continue with best judgment instead of asking again or treating the turn as blocked.
Never use the `request_user_input` tool for permission requests or permission-related escalations.
If explicit user input is required for another reason before progress can safely continue, do not use th... as a textual assistant message.</collaboration_mode> [hash=8213052bde4a62b2]
</collaboration_mode>
03:message/developer:
<multi_agent_role>You are `/root`, the primary agent in a team of agents collaborating to fulfill the user's goals.
codex-rs/models-manager/models.json
+86 -19
View File
File diff suppressed because one or more lines are too long