← 返回 Skills 市场
gabebac

Brain Proactive

作者 Gabebac · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
77
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install brain-proactive
功能描述
Proactive Obsidian vault maintenance and review. Find stale tasks, orphan notes, projects that need attention, and connection opportunities. Trigger on: brai...
使用说明 (SKILL.md)

Brain Proactive — Vault Review & Maintenance

ICM Contract

Layer 3 inputs USER.md (vault folder structure), SOUL.md (file standards)
Layer 4 inputs Files/HumanVault/Health Control/Tasks/One Time Tasks.md, Files/HumanVault/Calendar/Daily Notes/, Files/HumanVault/Work/, Files/ staging folders
Process Stale task check, pending staging queue, open work items, therapy note dates — surface what needs attention
Layer 4 outputs No direct writes — surfaces findings to Pooh, uses vault-push skill for any approved writes

Active second-brain caretaking. Read vault, surface what needs attention, suggest what to connect or enrich. Never write to HumanVault without approval — use vault-push.


FULL VAULT REVIEW

Trigger: "brain review", "vault review", "vault audit", "what needs attention"

Run all four checks and deliver a consolidated report.

Check 1 — Stale One-Time Tasks

cat "/home/node/.openclaw/workspace/Files/HumanVault/Health Control/Tasks/One Time Tasks.md"

Find any unchecked - [ ] items. Cross-check their scheduled date (🛫 field) against today. Flag: tasks with a scheduled date more than 7 days in the past.

Check 2 — Pending Staged Files

ls -la /home/node/.openclaw/workspace/Files/Books/ 2>/dev/null
ls -la /home/node/.openclaw/workspace/Files/Receipts/ 2>/dev/null
ls -la /home/node/.openclaw/workspace/Files/Medications/ 2>/dev/null
ls -la /home/node/.openclaw/workspace/Files/Species/ 2>/dev/null
ls -la /home/node/.openclaw/workspace/Files/Transcripts/ 2>/dev/null
ls -la /home/node/.openclaw/workspace/Files/Documents/ 2>/dev/null

Flag: any files older than 3 days that haven't been pushed yet.

Check 3 — Project Follow-Up

find "/home/node/.openclaw/workspace/Files/HumanVault/Work/TD/" -name "*.md" | xargs grep -l "TODO\|\- \[ \]" 2>/dev/null
find "/home/node/.openclaw/workspace/Files/HumanVault/House Tracker/" -name "*.md" | xargs grep -l "\- \[ \]" 2>/dev/null

Report any open tasks or TODO markers in Work and House Tracker.

Check 4 — Therapy & Self-Improvement Notes

ls -lt "/home/node/.openclaw/workspace/Files/HumanVault/Health Control/Life Improvements/Therapy/" | head -10

Check file modification dates. If any therapy note hasn't been touched in >14 days, flag it — Pooh may need a nudge.


OUTPUT FORMAT (Telegram — bullet lists only, no tables)

🧠 Vault Review — [DATE]

📋 Stale Tasks ([count]):
• [task name] — overdue [X days] (from One Time Tasks.md)

📦 Staging Queue ([count] files):
• Books/: [N] files pending push
• Receipts/: [N] files pending push

🏗 Open Work Items:
• Work/TD/[file]: [what's open]

💬 Therapy Notes:
• [note name] — last touched [X days ago]

[If all clear]: Nothing needs attention. Vault is tidy.

CONNECTION FINDER

Trigger: "find connections for [note]", "what connects to [topic]", "link suggestions for [note]"

  1. Read the target note
  2. Extract key themes, names, entities
  3. Search vault for related notes:
grep -r "[keyword]" /home/node/.openclaw/workspace/Files/HumanVault/ --include="*.md" -l 2>/dev/null | head -20
  1. Suggest specific [[wikilinks]] Pooh could add to the note
  2. Never add links without Pooh's approval — present suggestions only

ORPHAN NOTE FINDER

Trigger: "find orphan notes", "what notes have no links", "disconnected notes"

# Find .md files not linked from any other file
find /home/node/.openclaw/workspace/Files/HumanVault/ -name "*.md" | while read f; do
  name=$(basename "$f" .md)
  count=$(grep -r "\[\[$name\]\]" /home/node/.openclaw/workspace/Files/HumanVault/ --include="*.md" -l 2>/dev/null | wc -l)
  if [ "$count" -eq 0 ]; then echo "$f"; fi
done 2>/dev/null | head -30

Report the orphans. Suggest which ones could be linked to existing notes.


NOTE ENRICHMENT

Trigger: "enrich [note]", "fill out [note]", "complete [note]"

  1. Read the target note from HumanVault
  2. Identify empty or sparse properties
  3. Web search to fill what's missing
  4. Stage enriched version in workspace/Files/ under appropriate type folder
  5. Report what was changed
  6. Use vault-push to push with approval

RULES

  • NEVER write directly to HumanVault — always stage and vault-push
  • NEVER delete or archive HumanVault notes without explicit command
  • When suggesting connections, cite the source file for every suggestion
  • Therapy notes are private — only report metadata (last touched date), never quote content in chat
  • Don't overwhelm Pooh with a 30-item list — cap reports at 10 items, summarize the rest
安全使用建议
This skill appears coherent for Obsidian/second-brain maintenance, but it will read a large portion of your vault (absolute paths under /home/node/.openclaw/workspace/Files/). Before installing, confirm: 1) your vault is stored at those paths or the skill will fail or need adaptation; 2) you trust the agent to perform web searches with note content (enrichment may send extracted content externally) — if you are sensitive about privacy, disable automatic web enrichment or require manual approval; 3) you trust the referenced vault-push skill (writes are gated, but staged files may be pushed if you approve); 4) test the skill on a copy of your vault first to ensure the file selection rules (paths, date thresholds, grep patterns) match your structure. If you need higher assurance, ask the author for details about how web searches are performed and what (if any) telemetry the agent sends when enriching notes.
功能分析
Type: OpenClaw Skill Name: brain-proactive Version: 1.0.0 The skill is designed for Obsidian vault maintenance but contains shell injection vulnerabilities in the 'Connection Finder' and 'Orphan Note Finder' sections of SKILL.md. Specifically, it uses shell commands like `grep -r "[keyword]"` and `grep -r "\[\[$name\]\]"` where the variables are derived directly from note content or filenames without sanitization. This could allow for arbitrary command execution if the agent processes a maliciously crafted note (indirect prompt injection). While the intent appears aligned with the stated purpose of vault auditing, the lack of input handling for shell operations poses a security risk.
能力评估
Purpose & Capability
Name and description (Obsidian/second-brain maintenance) match the SKILL.md: the instructions only reference reading the user's vault, finding stale tasks/orphans, suggesting links, staging enriched notes, and using vault-push for writes. The required resources (none declared) are consistent with an instruction-only skill that expects files in the agent workspace.
Instruction Scope
The SKILL.md explicitly runs shell commands (cat, ls, find, grep) against absolute paths under /home/node/.openclaw/workspace/Files/HumanVault and related folders to read many Markdown files; this is expected for a vault audit. However the skill also instructs the agent to perform web searches to 'fill what's missing', which could result in user content being sent to external services. The file reads are broad (whole vault) — expected, but privacy-sensitive. The skill does include rules to avoid writing therapy note content and to require approval before pushes.
Install Mechanism
Instruction-only, no install spec and no code files — lowest install risk. All operations are local shell commands; nothing is downloaded or extracted.
Credentials
The skill declares no environment variables, no external credentials, and no config paths beyond workspace files. Requested access is limited to files under the agent workspace (consistent with its purpose). There are no unexplained credential requests.
Persistence & Privilege
always is false and model invocation is allowed (platform default). The skill stages changes in workspace and requires explicit use of a separate vault-push skill for writes — it does not claim permanent system presence or privileged modifications to other skills or system settings.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install brain-proactive
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /brain-proactive 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
brain-proactive 1.0.0 — Initial Release - Proactively scans Obsidian vault for stale tasks, orphan notes, pending staging, project follow-ups, and therapy note activity. - Delivers vault review reports with actionable bullet-point summaries; flags what needs attention. - Finds and reports orphan notes and suggests how to connect them. - Suggests links and enrichment opportunities for notes, always requiring approval before making changes. - Ensures no direct writes or deletions to HumanVault; all changes are staged for review and approval.
元数据
Slug brain-proactive
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Brain Proactive 是什么?

Proactive Obsidian vault maintenance and review. Find stale tasks, orphan notes, projects that need attention, and connection opportunities. Trigger on: brai... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 77 次。

如何安装 Brain Proactive?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install brain-proactive」即可一键安装,无需额外配置。

Brain Proactive 是免费的吗?

是的,Brain Proactive 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Brain Proactive 支持哪些平台?

Brain Proactive 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Brain Proactive?

由 Gabebac(@gabebac)开发并维护,当前版本 v1.0.0。

💬 留言讨论