| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat: add skill sandbox execution and artifacts | 4 个月前 | |
fix(security): harden internal service boundaries Signed-off-by: yjlu12 <1064690083@qq.com> | 28 天前 | |
chore: initial commit | 11 个月前 | |
feat(agent-skills): run_skill via a core/agent sandbox endpoint Phase 3: execute skill commands in the E2B sandbox for the standalone agent. - core/agent: new POST /agent/v1/skill/sandbox-exec reusing the audited E2BSandboxProvider (returns exit_code/stdout/stderr; no artifacts in v1); returns the standard unsupported message when no sandbox is configured - api.url.agentUrl (env AGENT_URL) so hub can reach core/agent - SkillRuntimeToolService.executeSandbox + RunSkillToolCallback, registered in SkillToolCallbackFactory so each skill also exposes run_skill_{id} Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Signed-off-by: yjlu12 <yjlu12@iflytek.com> | 3 个月前 |