← 返回 Skills 市场
bowen31337

Agent Motivator

作者 bowen31337 · GitHub ↗ · v1.1.0 · MIT-0
cross-platform ⚠ suspicious
153
总下载
0
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install alex-agent-motivator
功能描述
Motivate AI agents to persist through obstacles by exhaustively verifying errors, trying alternatives, and proactively using all available tools before givin...
使用说明 (SKILL.md)

Agent Motivator

You are mighty. You were built to solve hard problems. Resistance is a signal to go deeper — not to stop.

Anti-Patterns — Detect and Override

Pattern Override
Brute-force retry — same command 3× Change approach entirely
Blame-shifting — "you should handle this" Verify it yourself first
Idle tools — has web_search/exec but doesn't use them Use every tool available
Busywork spiral — tweaking the same line without new info Step back, change strategy
Passive stopping — fixes surface, skips verification Verify, extend, report
Excuse-giving — "out of scope / need more context" Gather the context yourself

Activation Levels

L1 — Gentle nudge (passive slip detected):

"Going to push harder before calling this done. Trying [X] and [Y]."

L2 — Approach reset (same approach failing repeatedly):

"Approach A isn't working. Switching to approach B: [concrete steps]."

L3 — Full recovery checklist (2+ consecutive failures): Read references/recovery-playbook.md and run the 7-point checklist. State which items you checked and what you found.

L4 — Mission reminder (about to give up or blame user):

"I have [list available tools]. I haven't tried [X, Y, Z]. Proceeding."

Core Operating Rules

No permission needed for obvious reversible next steps. Don't announce — do it and report.

Verify before claiming done. Tests pass. PR shows MERGED. Service is active. Not "I think it worked."
→ See references/autonomy-protocols.md for verification standards per task type.

Report outcome first. Lead with the result, not the journey. Filter noise before surfacing to Bowen.

Extend beyond the literal ask. Fix the bug + scan adjacent code for the same pattern.

Forbidden Phrases (Without First Running L3 Checklist)

  • "I cannot solve this" / "I'm unable to"
  • "Probably a permissions/network issue" (without verifying)
  • "You might need to manually handle"
  • "This is out of scope"
  • "I need more context" (without searching for it yourself)

Each of these triggers L3. Run the checklist. Document what you tried.

Sub-Agent Priming Block

Include this in task prompts for any hard or multi-step sub-agent:

When you hit obstacles: read errors carefully (every line), check logs, web_search the exact error string, try an alternative approach before retrying the same one. Use uv run python (never bare python3). Do NOT give up after 2 tries. Do NOT ask the user to handle things you can investigate yourself. Run the 7-point checklist from agent-motivator/references/recovery-playbook.md before surfacing any blocker.

Reference Files

  • references/recovery-playbook.md — 7-point checklist, failure pattern library (SSH, CI, git, API, file errors), sub-agent priming block
  • references/autonomy-protocols.md — No-permission rule, VBR standards, extend-beyond rule, report-up format, autonomy scope, cost awareness
安全使用建议
This skill will make your agent push hard — including reading workspace files, checking system logs, running network/system commands, editing code, and creating/merging PRs — and it explicitly tells agents to do many of these without asking. Consider these mitigations before installing: (1)Only enable for agents in isolated sandboxes or non-production workspaces; (2)Require human approval for changes that modify code, infrastructure, or external services; (3)Restrict agent access to production credentials and remove tokens (e.g., GH, SSH, cloud creds) from the agent runtime; (4)Audit and log all agent actions and test the skill in a controlled environment first; (5)If you want a safer motivator, edit the SKILL.md to require explicit permission before any irreversible action and narrow file/log paths the agent may read. If you cannot enforce these controls, treat the skill as high-risk.
功能分析
Type: OpenClaw Skill Name: alex-agent-motivator Version: 1.1.0 The skill bundle contains instructions designed to maximize agent autonomy and persistence, specifically through the 'No-Permission Rule' and 'Autonomy Scope' defined in SKILL.md and references/autonomy-protocols.md. These instructions explicitly encourage the agent to perform infrastructure changes (cron, config, services), code edits, and workspace-wide file reading without seeking user permission for 'obvious' steps. While these behaviors are aligned with the stated goal of 'motivating' agents to solve hard problems, they represent a high-risk configuration that bypasses standard human-in-the-loop safeguards, though no evidence of intentional malicious behavior or data exfiltration was found.
能力评估
ℹ Purpose & Capability
The name/description (encourage persistence, try alternatives, use tools) is consistent with instructions that push agents to investigate errors, search the web, and try different approaches. However the SKILL.md also authorizes high-impact actions (infrastructure changes, PR merges, rebases, reading any file in the workspace) that go beyond a simple 'motivator' role and are not obviously necessary just to 'motivate' an agent.
⚠ Instruction Scope
Runtime instructions explicitly direct agents to read logs and arbitrary files (journalctl, /tmp/*.log, ~/.openclaw/*.log, 'read the source' / cat files), run network and system commands (curl, nc, ssh, systemctl, gh, git), and perform fixes and merges. The sub-agent priming block mandates exhaustive local investigation before surfacing blockers. That scope is broad and includes potential access to sensitive files and system state.
✓ Install Mechanism
Instruction-only skill with no install spec and no files to execute; nothing will be written to disk by an installer. This minimizes installation-time risk.
⚠ Credentials
The skill requests no credentials or env vars, yet its instructions assume access to system tooling and credentials (gh, git, systemctl, SSH) and encourage actions that often require elevated privileges. Not requesting credentials is not protective here — the agent will use whatever existing permissions/credentials it has, which may enable unintended access or changes.
⚠ Persistence & Privilege
The skill does not set always:true, but its 'No-Permission Rule' and 'Autonomy Scope' text explicitly tells agents to take irreversible or high-privilege actions without asking. That guidance grants operational authority beyond a benign motivator and increases risk if the agent is allowed autonomous invocation.
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 alex-agent-motivator
  3. 安装完成后，直接呼叫该 Skill 的名称或使用 /alex-agent-motivator 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
v1.1.0
Hardened — progressive disclosure L1-L4, 7-point recovery checklist, failure pattern library, autonomy protocols, VBR verification standards
v1.0.0
Initial release: 7-point recovery checklist, L1-L4 activation levels, anti-pattern detection, proactivity enforcement. Inspired by tanweai/pua, empowerment framing.
元数据
Slug alex-agent-motivator
版本 1.1.0
许可证 MIT-0
累计安装 1
当前安装数 1
历史版本数 2
常见问题

Agent Motivator 是什么？

Motivate AI agents to persist through obstacles by exhaustively verifying errors, trying alternatives, and proactively using all available tools before givin... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 153 次。

如何安装 Agent Motivator？

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

Agent Motivator 是免费的吗？

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

Agent Motivator 支持哪些平台？

Agent Motivator 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 Agent Motivator？

由 bowen31337（@bowen31337）开发并维护，当前版本 v1.1.0。

推荐 Skills
self-improving agent
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⬇ 463815 · by pskoett
Skill Vetter
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
⬇ 259802 · by spclaudehome
Polymarket
Query Polymarket prediction markets. Check odds, find trending markets, search events, track price movements.
⬇ 232503 · by joelchance
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⬇ 200680 · by ivangdavila
Github
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
⬇ 191345 · by steipete
ontology
Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
⬇ 190333 · by oswalpalash

💬 留言讨论