KYVAutoResearch-v2SAFE
Project-based auto-research AI agent framework. Problem × Evidence Matrix, evidence grading, MCP Server. Python 3.10+. | 项目制自动研究 Agent 框架:问题 × 证据矩阵、证据分级。
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
项目制的自动研究 Agent 框架。让 AI 围绕研究主题把证据穷尽到可以下判断的程度;证据不足时,如实说出来。 Project-based Auto-Research AI Agent Framework. Exhausts evidence until a judgment can be made; honestly states "insufficient evidence" when there isn't enough.
English | 中文
[](https://github.com/kyvjyf/KYVAutoResearch-v2/releases) [](https://www.python.org/) [](https://github.com/kyvjyf/KYVAutoResearch-v2/blob/main/LICENSE) [](https://modelcontextprotocol.io/)
📖 目录 | Table of Contents
- 项目简介 | Introduction
- 核心概念 | Core Concepts
- 快速开始 | Quick Start
- 项目结构 | Project Structure
- 设计文档 | Design Documentation
- 边界说明 | Notes
- 反馈 | Feedback
项目简介 | Introduction
KYVAutoResearch 是一个项目制的自动研究 Agent 框架。它不同于普通的对话式 AI,而是围绕研究主题建立问题树,用问题 × 证据矩阵追踪证据,按证据等级判定。它的核心承诺是:证据不足时,如实输出“无法判断”,绝不编造。
KYVAutoResearch is a project-based auto-research AI agent framework. Unlike standard conversational AI, it builds a problem tree around a research topic, tracks evidence using a Problem × Evidence Matrix, and makes judgments based on evidence levels. Its core promise: when evidence is insufficient, it honestly outputs "Unable to determine" instead of hallucinating.
核心概念 | Core Concepts
- 问题 × 证据矩阵 | Problem × Evidence Matrix:追踪每一条证据与问题的对应关系,确保没有无证据的结论。
- 双通道判定 | Dual-channel Judgment:确保证据充分且逻辑自洽。
- 证据分级与缺口识别 | Evidence Grading & Gap Identification:自动识别证据不足的环节,拒绝在覆盖率未满 100% 时给出“已解决”的建议。
- MCP Server + Dual Skill:包含
KYVAutoResearch2MCP,支持与 Claude Desktop 等客户端无缝集成,并配有双 Skill 设计。
快速开始 | Quick Start
b2bab39557c6OBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add kvy-autoresearch-v2-mcp -- uvx kvy-autoresearch-v2-mcp
{
"mcpServers": {
"kvy-autoresearch-v2-mcp": {
"command": "uvx",
"args": [
"kvy-autoresearch-v2-mcp"
]
}
}
}Exposed tools (64)
49 read · 15 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
add_keywords | write | |
add_question_dependency | write | |
add_sub_question | write | |
approve_question_tree | read | |
check_url_seen | read | |
confirm_answered | read | |
create_human_request | write | |
create_project | write | |
decompose_questions | read | |
end_research_session | read | |
exclude_document | read | |
exclude_evidence_item | read | |
export_dataset | read | |
export_matrix | read | |
export_question_tree | read | |
export_report | read | |
get_coverage | read | |
get_document_excerpt_by_keyword | read | |
get_event_log | read | |
get_evidence_item | read | |
get_followup_chain | read | |
get_keyword_bank | read | |
get_matrix_view | read | |
get_next_actions | read | |
get_pending_cells | read | |
get_pending_requests | read | |
get_project_info | read | |
get_question_detail | read | |
get_question_tree | read | |
get_questions_by_status | read | |
get_request_detail | read | |
get_research_status | read | |
get_round_report | read | |
get_schema_info | read | |
get_search_history | read | |
get_search_stats | read | |
get_storage_stats | read | |
get_utilization | read | |
input_followup_question | read | |
input_root_question | read | |
list_documents | read | |
list_evidence_items | read | |
list_projects | read | |
list_unanswered_questions | read | |
log_search | read | |
mark_needs_more | read | |
override_cell | read | |
read_document_excerpt | read | |
register_document | read | |
resolve_request | read | |
review_root_decision | read | |
save_cell_verdict | write | |
save_evidence_item | write | |
save_planner_action | write | |
search_documents_fts | read | |
search_evidence_fts | read | |
set_stage_sufficient | write | |
start_research_session | write | |
submit_request_batch | write | |
suggest_answered | read | |
update_evidence_item | write | |
update_keyword_score | write | |
update_project_config | write | |
update_question | write |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | WARN |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- declared (2 observation(s))
- Shell
- declared (1 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (2)
modules_cache[alias] = importlib.import_module(mod_name)
fastmcp
Gates applied: no_behavioural_pass.
b2bab39557c6full audit observations/trust-audit/mcp-server/kyvjyf__kyvautoresearch-v2.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | b2bab39557c6 | SAFE | B | 89 | first audit |
Questions
What is the KYVAutoResearch-v2 MCP server?
Project-based auto-research AI agent framework. Problem × Evidence Matrix, evidence grading, MCP Server. Python 3.10+. | 项目制自动研究 Agent 框架:问题 × 证据矩阵、证据分级。
What tools does KYVAutoResearch-v2 expose?
64 in total: 49 read-only, 15 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is KYVAutoResearch-v2 safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does KYVAutoResearch-v2 need?
No credential environment variables were found in its source, so it appears to need none.
How does KYVAutoResearch-v2 run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as kvy-autoresearch-v2-mcp.
How current is this page?
The grade is for one exact copy of the source (b2bab39557c6), read on 2026-10-11. The repository is watched and re-audited when it changes.