| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat(session): enhance session management with backend instance support This commit introduces significant updates to the session management system, including the addition of a backend instance to the SessionRecord struct and the implementation of backend leasing and synchronization in the CoreBackedSessionService. The changes ensure that session instances are properly managed and persisted, enhancing the overall functionality and reliability of session operations. Additionally, new tests have been added to validate the persistence of active backend instances during session operations. | 30 天前 | |
Move conch backend ownership to gateway | 1 个月前 | |
fix(tui): input auto-wrap, Alt+Enter newline, and intelligent arrow key navigation By the way, apply cargo fmt for code style consistency. Co-Authored-By: GLM-5 | 19 天前 | |
fix(tui): input auto-wrap, Alt+Enter newline, and intelligent arrow key navigation By the way, apply cargo fmt for code style consistency. Co-Authored-By: GLM-5 | 19 天前 | |
cargo fmt | 30 天前 | |
fix(subagent): apply slim read-only profile only to default subagents The slim exploration profile (read-only tool allowlist, slim system prompt, no skills) was keyed on is_subagent (agent_id != main), so it also clobbered custom-role subagents — which carry their own prompt and tool list and must keep them. Thread subagent_role_id from spawn (SessionAgentRecord) through runtime_input_from_session into SessionRuntimeBuildInput, and gate the profile on is_default_subagent = is_subagent && role_id.is_none(). Custom-role subagents now fall back to the composed prompt + full toolset + skills (no repo map / delegation rules, which stay main-agent-only). build_tool_registry takes an explicit apply_readonly_profile flag instead of recomputing is_subagent. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> | 17 天前 | |
feat(subagent): slim read-only subagent profile, turn budget, static repo map - Subagents get a dedicated slim system prompt and are restricted to read-only search tools (file_read/glob/grep), enforced even under an explicit --tools allowlist; skills are withheld from the child. - Caller-settable turn budget with a larger exploration default (~20); a fan-out prompt issues independent grep/glob/file_read in one turn. - A bounded static repository map is injected once into the MAIN agent's cache-stable prefix; subagents are excluded. - The subagent prompt gains a "disposable exploration worker" distillation instruction; the coordinator prompt-builder tests move to substring checks. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> | 18 天前 | |
refactor(gateway): remove unused Arc import and update trace_id variable - Removed the unused Arc import from pending_interaction.rs. - Updated the trace_id variable in trace_recorder.rs to be prefixed with an underscore, indicating it is intentionally unused. | 2 个月前 | |
feat(sandbox): introduce /allow-path command for granting sandbox path permissions - Added a new TUI command /allow-path to allow users to grant read or write permissions for specific paths in the local sandbox. - Implemented functionality to handle path resolution and permission granting within the GatewayRuntime. - Enhanced the permission control interface to support sandbox denial checks for specified paths. - Updated documentation to reflect the new command and its usage. This feature improves user control over sandbox permissions, allowing for more flexible interaction with the local environment. | 27 天前 | |
fix: resolve sync conflicts after updating master | 2 个月前 | |
test: repair pre-existing drifted/stale test fixtures on dev The hook and app test trees fail to COMPILE on dev, and two subagent coordinator tests fail at runtime — all pre-existing drift where production behavior already landed but the test fixtures/expectations were never updated. This realigns them so cargo test --workspace is green again; land it first so the test suites compile and pass for every other branch. - crates/hook/.../tool/adaptor.rs — display_value fixture - apps/xiaoo-app/.../tui/state/app_state.rs — is_secret field + cycle test - apps/xiaoo-app/.../gateway/prompt_utils.rs — Subagent Delegation expectation - crates/subagent/src/coordinator.rs — subagent_prompt_builder_{with,without}_schema expected the old "xiaoo-guardian" preamble the template no longer emits; corrected to the current "For general conversations…" preamble. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> | 19 天前 | |
feat: implement pending user message handling in chat state and gateway - Added support for managing pending user messages within the chat state and gateway runtime. - Introduced methods for enqueueing and discarding pending user messages, enhancing user interaction during ongoing tasks. - Updated the agent loop to drain pending user messages and integrate them into the chat flow. - Enhanced session management to accommodate pending messages, ensuring a smoother user experience. - Updated relevant tests to verify the functionality of pending user message handling. | 1 个月前 | |
feat(sandbox): introduce /allow-path command for granting sandbox path permissions - Added a new TUI command /allow-path to allow users to grant read or write permissions for specific paths in the local sandbox. - Implemented functionality to handle path resolution and permission granting within the GatewayRuntime. - Enhanced the permission control interface to support sandbox denial checks for specified paths. - Updated documentation to reflect the new command and its usage. This feature improves user control over sandbox permissions, allowing for more flexible interaction with the local environment. | 27 天前 | |
fix(subagent): apply slim read-only profile only to default subagents The slim exploration profile (read-only tool allowlist, slim system prompt, no skills) was keyed on is_subagent (agent_id != main), so it also clobbered custom-role subagents — which carry their own prompt and tool list and must keep them. Thread subagent_role_id from spawn (SessionAgentRecord) through runtime_input_from_session into SessionRuntimeBuildInput, and gate the profile on is_default_subagent = is_subagent && role_id.is_none(). Custom-role subagents now fall back to the composed prompt + full toolset + skills (no repo map / delegation rules, which stay main-agent-only). build_tool_registry takes an explicit apply_readonly_profile flag instead of recomputing is_subagent. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> | 17 天前 | |
feat(session): enhance session management with backend instance support This commit introduces significant updates to the session management system, including the addition of a backend instance to the SessionRecord struct and the implementation of backend leasing and synchronization in the CoreBackedSessionService. The changes ensure that session instances are properly managed and persisted, enhancing the overall functionality and reliability of session operations. Additionally, new tests have been added to validate the persistence of active backend instances during session operations. | 30 天前 | |
feat: initial commit with core features | 2 个月前 | |
feat: add reasoning effort configuration to CLI and related components - Introduced ReasoningEffort enum to manage reasoning effort levels (Off, High, Max). - Updated CLI commands to accept reasoning_effort as an argument. - Enhanced configuration structures and session management to include reasoning effort. - Modified various components to utilize the new reasoning effort settings, ensuring consistent behavior across the application. - Added default values and serialization support for reasoning effort in relevant structs. | 1 个月前 | |
fix(subagent): apply slim read-only profile only to default subagents The slim exploration profile (read-only tool allowlist, slim system prompt, no skills) was keyed on is_subagent (agent_id != main), so it also clobbered custom-role subagents — which carry their own prompt and tool list and must keep them. Thread subagent_role_id from spawn (SessionAgentRecord) through runtime_input_from_session into SessionRuntimeBuildInput, and gate the profile on is_default_subagent = is_subagent && role_id.is_none(). Custom-role subagents now fall back to the composed prompt + full toolset + skills (no repo map / delegation rules, which stay main-agent-only). build_tool_registry takes an explicit apply_readonly_profile flag instead of recomputing is_subagent. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> | 17 天前 | |
fix(tui): fix context loss by saving state on LLM provider errors Co-Authored-By: GLM-5 | 17 天前 | |
feat: initial commit with core features | 2 个月前 | |
| 15 天前 | ||
fix(tui): fix context loss by saving state on LLM provider errors Co-Authored-By: GLM-5 | 17 天前 | |
feat(tool/bash): add interactive command detection to prevent execution requiring user interaction Co-Authored-By: GLM-5 | 20 天前 | |
feat: add reasoning effort configuration to CLI and related components - Introduced ReasoningEffort enum to manage reasoning effort levels (Off, High, Max). - Updated CLI commands to accept reasoning_effort as an argument. - Enhanced configuration structures and session management to include reasoning effort. - Modified various components to utilize the new reasoning effort settings, ensuring consistent behavior across the application. - Added default values and serialization support for reasoning effort in relevant structs. | 1 个月前 | |
tui: fix context usage display 1. Split Tok and Ctx semantics so Tok remains total token usage while Ctx shows input context usage against the configured context window. 2. Carry estimated_input_tokens from core agent outcomes through the gateway into the TUI status panel. 3. Fall back to prompt_tokens when estimated input usage is unavailable and render Ctx as used/limit. Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com> | 2 个月前 | |
feat(subagent): slim read-only subagent profile, turn budget, static repo map - Subagents get a dedicated slim system prompt and are restricted to read-only search tools (file_read/glob/grep), enforced even under an explicit --tools allowlist; skills are withheld from the child. - Caller-settable turn budget with a larger exploration default (~20); a fan-out prompt issues independent grep/glob/file_read in one turn. - A bounded static repository map is injected once into the MAIN agent's cache-stable prefix; subagents are excluded. - The subagent prompt gains a "disposable exploration worker" distillation instruction; the coordinator prompt-builder tests move to substring checks. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> | 18 天前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 30 天前 | ||
| 1 个月前 | ||
| 19 天前 | ||
| 19 天前 | ||
| 30 天前 | ||
| 17 天前 | ||
| 18 天前 | ||
| 2 个月前 | ||
| 27 天前 | ||
| 2 个月前 | ||
| 19 天前 | ||
| 1 个月前 | ||
| 27 天前 | ||
| 17 天前 | ||
| 30 天前 | ||
| 2 个月前 | ||
| 1 个月前 | ||
| 17 天前 | ||
| 17 天前 | ||
| 2 个月前 | ||
| 15 天前 | ||
| 17 天前 | ||
| 20 天前 | ||
| 1 个月前 | ||
| 2 个月前 | ||
| 18 天前 |