Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Pulse TODO
v0.3.0Unified task management and scheduling for AI agents. Use when: (1) a commitment is made (I'll do X, 帮你跟进, remember to), (2) checking what's pending (待办, wha...
⭐ 0· 74·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
The skill describes a local TODO/cron-based task manager (reasonable). However, the SKILL.md and setup.md explicitly instruct the agent to read/write TODO.md in the workspace root and to update OpenClaw cron jobs (~/.openclaw/cron/jobs.json). The registry metadata declared no required config paths, binaries, or credentials. That mismatch means the skill expects filesystem and platform-level access that isn't declared.
Instruction Scope
Runtime instructions tell the agent to: add/modify TODO.md, create/disable cron jobs, edit HEARTBEAT.md, and migrate existing cron jobs and memory files. These are concrete filesystem and platform operations (including modifying ~/.openclaw/cron/jobs.json). The SKILL.md also expects the agent to message humans for nudges, but it does not specify which channel or required credentials. Instructions therefore reach into system state and platform config beyond a simple read-only helper.
Install Mechanism
This is an instruction-only skill with no install spec and no code files to execute. That minimizes supply-chain risk because nothing is downloaded or installed by the skill package itself.
Credentials
The manifest lists no required env vars or config paths, yet the instructions require filesystem access (workspace TODO.md, ~/.openclaw/cron/jobs.json, HEARTBEAT.md, memory files). This is disproportionate: either the manifest should declare these config paths/permissions, or the instructions should avoid demanding them. The skill also implies sending messages (nudging humans) but doesn't declare what credentials or channels are needed.
Persistence & Privilege
always:false and normal autonomous invocation are fine. The bigger concern is that the skill's workflow explicitly instructs modifying platform-level cron configurations (OpenClaw cron jobs), which is a system-wide change affecting scheduled behavior. That level of effect should be declared and approved; the skill does not declare or constrain those changes.
What to consider before installing
This skill appears to be a local TODO + cron manager and contains explicit instructions to modify TODO.md and OpenClaw cron jobs (e.g., ~/.openclaw/cron/jobs.json) even though the manifest doesn't declare those config paths or credentials. Before installing or enabling this skill you should:
- Confirm whether your agent runtime permits skills to modify files under ~/.openclaw and create/disable cron jobs. If you don't want that, do not enable the skill or run it only in a sandboxed workspace.
- Back up your current TODO.md, HEARTBEAT.md, and ~/.openclaw/cron/jobs.json so you can revert changes.
- Ask the skill author (or inspect additional docs) which exact cron job names and commands the skill will create, and whether it will overwrite or remove existing cron entries. Require a dry-run mode or an explicit approval step for creating/deleting cron jobs.
- Verify how the skill 'nudges' humans: what channel (chat, email, notifications) will be used and what credentials/tokens are required. If messaging requires tokens, prefer creating a dedicated, limited-scope service account rather than reusing platform-wide credentials.
- If you want lower risk, run pulse-todo in a dedicated workspace directory and restrict file permissions so it cannot touch other projects or global OpenClaw config.
Given the manifest/instruction mismatch, do not assume this skill only reads data — it instructs writes and platform modifications. If you cannot confirm the above details, treat installation as potentially disruptive and enable only after explicit review or sandboxed testing.Like a lobster shell, security has layers — review code before you run it.
latestvk97ej8wnvm3dbche2m1y059py583zs7f
License
MIT-0
Free to use, modify, and redistribute. No attribution required.