Evidence Anchor

v2.0.0

标准化定义和验收项目证据锚点,确保长期记忆可验证、可追溯、可复用,支持多层级证据及定期复核流程。

0· 66·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for jiangwill2023/evidence-anchor.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Evidence Anchor" (jiangwill2023/evidence-anchor) from ClawHub.
Skill page: https://clawhub.ai/jiangwill2023/evidence-anchor
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install evidence-anchor

ClawHub CLI

Package manager switcher

npx clawhub@latest install evidence-anchor
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
The name/description (standardizing evidence anchors for verifiable, traceable memory) aligns with the SKILL.md content: templates and step-by-step verification for URLs, file paths, configs, logs, commits, timestamps, etc. The examples and acceptance criteria are consistent with the stated purpose.
Instruction Scope
The SKILL.md explicitly tells an agent to verify evidence by actions such as curl-ing endpoints, SSH-ing into servers, reading files/configs, and writing evidence into memory. Those actions are appropriate for the skill's goal, but they require access to systems and may expose sensitive data. The skill does not itself enumerate how those accesses/credentials are supplied or any limits on what to read or transmit.
Install Mechanism
This is an instruction-only skill with no install steps, binaries, or code to download. No install-related risk is present.
Credentials
The skill declares no required environment variables or credentials, yet many verification steps imply needing SSH keys, service credentials, or network access to private URLs and files (e.g., bank statements, /opt/ paths, /etc configs). That omission is not necessarily malicious but is an important operational gap: the user must ensure appropriate and least-privilege access is provided and that sensitive artifacts are handled carefully.
Persistence & Privilege
Flags are default (not always:true). The skill does instruct writing standardized evidence into memory (explicit part of the workflow), which is expected for this purpose. Autonomous invocation is allowed by platform default; there is no special elevated persistence requested by the skill itself.
Assessment
This skill is a documentation and verification template (benign and coherent), but before installing or using it: 1) Confirm where and how the agent will obtain SSH credentials, API keys, or file system access — do not place sensitive keys into an unreviewed environment. 2) Limit the agent's permissions (use least privilege) so it can verify only the systems/paths you expect. 3) Audit what the agent will write into its memory—avoid storing raw sensitive artifacts (bank statements, private keys); prefer redacted summaries or hashes. 4) Test the workflow in a safe environment first (non-production server) to ensure it performs only expected checks (curl/ls/cat) and does not transmit data to unexpected endpoints. 5) If you require stronger guarantees, ask the maintainer to declare required credentials and add explicit guidance about redaction, retention, and outbound transmission policies.

Like a lobster shell, security has layers — review code before you run it.

latestvk970c8pp9g17x833gnh7meg5c185jqhe
66downloads
0stars
2versions
Updated 1d ago
v2.0.0
MIT-0

evidence-anchor Skill

版本: v1.0
创建日期: 2026-04-26
来源: Agent 记忆补强两层方法论验证


用途

标准化定义和验收"证据锚点",确保长期记忆不是"口头记忆",而是可验证、可回溯、可复用的资产。


适用场景

  • 项目记忆需要沉淀证据
  • 结果闭环型记忆写入
  • 跨 agent 交接时需要统一证据口径
  • 定期复盘时需要验证历史结论

证据分级

Level 1:直接证据(最强)

定义: 可直接验证项目状态/结果的原始证据。

示例:

类型示例
线上验证https://example.com/ 返回 200 OK
文件路径/opt/xxx/app/current/index.html
配置路径/etc/nginx/conf.d/xxx.conf
Commit hashabc123def
CI 日志GitHub Actions Run #456
部署时间戳YYYY-MM-DD HH:MM TZ
签署文件projects/xxx-contract-signed.pdf
银行流水bank-statement-YYYY-MM.pdf

验收标准:

  • 路径/URL 可直接访问或查验
  • 时间戳精确到日(最好到时分)
  • 文件真实存在且内容匹配

Level 2:间接证据(中等)

定义: 可佐证项目进展,但不能直接证明结果的证据。

示例:

类型示例
设计文档projects/xxx-design.md
会议纪要meetings/2026-04-16-xxx.md
方案草稿projects/xxx-plan-v0.md
工作卡xxx-work-card.md
基线文件xxx.baseline.js

验收标准:

  • 文档真实存在
  • 内容与所述项目相关
  • 有明确创建/修改日期

使用限制:

  • 不能单独支撑 DONE 状态
  • 可支撑 PARTIAL 状态(设计已完成,实施未启动)

Level 3:引用证据(最弱)

定义: 对他人陈述/记忆的引用,不能独立验证。

示例:

类型示例
记忆引用memory/2026-04-16.md 中的记录
口头确认"用户说已经完成了"
二手转述"听某某说部署好了"

验收标准:

  • 引用路径明确
  • 原始来源可追溯

使用限制:

  • 不能单独作为项目状态证据
  • 仅可作为辅助佐证
  • 优先追 Level 1/2 证据

证据锚点定义规范

标准格式

每条证据锚点应包含:

- **证据类型**:[线上验证/文件路径/配置路径/日志记录/签署文件/其他]
- **证据路径**:[具体路径或 URL]
- **验证方式**:[如何查验该证据]
- **证据结论**:[该证据支撑什么结论]

示例

示例 1:官网修复证据(脱敏版)

- **证据类型**:线上验证 + 文件路径
- **证据路径**:
  - `https://example.com/` 返回 200 OK
  - `/opt/xxx/app/current/index.html`
  - `/etc/nginx/conf.d/xxx.conf`
- **验证方式**:
  - curl 线上 URL
  - SSH 登录服务器查看文件
  - readback Nginx 配置
- **证据结论**:官网主链路已恢复,Nginx 配置已修复

示例 2:部署项目证据(脱敏版)

- **证据类型**:文件路径 + 健康检查
- **证据路径**:
  - `/opt/xxx/docker-compose.yml`
  - `/health` 端点返回 200
- **验证方式**:
  - SSH 查看部署文件
  - curl https://example.com/health
- **证据结论**:部署骨架已落地,健康检查可用

示例 3:融资项目证据(脱敏版)

- **证据类型**:文档路径 + 许可记录
- **证据路径**:
  - `projects/xxx-finance.md`
  - 许可记录(YYYY-MM-DD 取得)
- **验证方式**:
  - 读取项目文档
  - 查验许可编号/官方记录
- **证据结论**:融资方案已形成,许可已落地,银行贷款未见落地证据

证据验收流程

Step 1:识别证据类型

判断证据属于:
- Level 1(直接证据)
- Level 2(间接证据)
- Level 3(引用证据)

Step 2:验证证据有效性

检查:
- 路径/URL 是否真实存在
- 内容是否与所述匹配
- 日期是否在合理范围内
- 是否有篡改/过期风险

Step 3:判断证据支撑力

对照状态判断:
- DONE:需要 Level 1 证据支撑关键阶段
- PARTIAL:Level 1 + Level 2 混合
- BLOCKED:只有 Level 2/3 或无证据

Step 4:写入记忆

按标准格式写入:
- 证据类型
- 证据路径
- 验证方式
- 证据结论

Step 5:定期复核

建议每 1-3 个月复核:
- 线上 URL 是否仍可达
- 文件是否仍存在于路径
- 配置是否仍生效
- 是否有新证据可补充

证据与状态映射

项目状态必需证据可选证据
DONELevel 1 证据支撑所有关键阶段Level 2 辅助说明设计/背景
PARTIALLevel 1 证据支撑已落地阶段 + Level 2 说明未落地阶段Level 3 引用记忆
BLOCKEDLevel 2 证明任务已接收 + Level 3 说明阻塞原因-

常见陷阱

陷阱 1:只有文档没有实施证据

表现: 只有 xxx-plan.md,没有部署/签署/上线证据

解法: 明确标注"设计已完成,实施未启动",状态最多报 PARTIAL

陷阱 2:证据路径模糊

表现: "服务器上有"、"应该部署了"

解法: 强制要求具体路径,如 /opt/xxx/app/current

陷阱 3:用 Level 3 撑 DONE

表现: 只有"听某某说完成了",没有 Level 1 证据

解法: 降级为 PARTIAL/BLOCKED,直到拿到 Level 1 证据

陷阱 4:证据过期未更新

表现: URL 已 404、文件已删除、配置已覆盖

解法: 定期复核,证据失效时同步更新记忆状态


证据锚点模板

模板 1:线上服务证据

- **证据类型**:线上验证
- **证据路径**:`https://[domain]/[path]`
- **验证方式**:curl / browser 访问
- **预期结果**:返回 [状态码] / 显示 [内容]
- **证据结论**:[支撑什么结论]

模板 2:服务器文件证据

- **证据类型**:文件路径
- **证据路径**:`/path/to/file`
- **验证方式**:SSH 登录 + cat/ls
- **预期结果**:文件存在,内容包含 [关键信息]
- **证据结论**:[支撑什么结论]

模板 3:配置证据

- **证据类型**:配置路径
- **证据路径**:`/path/to/config.conf`
- **验证方式**:readback 配置文件
- **预期结果**:配置包含 [关键规则]
- **证据结论**:[支撑什么结论]

模板 4:日志证据

- **证据类型**:日志记录
- **证据路径**:`memory/YYYY-MM-DD.md` 或 `logs/xxx.log`
- **验证方式**:读取日志文件
- **预期结果**:日志包含 [关键事件/时间戳]
- **证据结论**:[支撑什么结论]

相关 Skill

  • memory-backfill - 记忆补强标准化流程
  • result-closure-memory - 结果闭环型记忆写入规范
  • taskflow - 任务流管理

维护者

  • 创建者:小强(qiang)
  • 创建日期:2026-04-26
  • 来源项目:Agent 记忆补强两层方法论验证

变更日志

版本日期变更内容
v1.02026-04-26初始版本,基于 4 位 agent 验证通过

Comments

Loading comments...