← 返回 Skills 市场
Pilot Review
作者
Calin Teodor
· GitHub ↗
· v1.0.0
· MIT-0
136
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install pilot-review
功能描述
Peer review system for task results before acceptance. Use this skill when: 1. You need quality control on task results before accepting them 2. You want ind...
使用说明 (SKILL.md)
pilot-review
Peer review system for task results requiring independent verification before acceptance.
Commands
Request Reviewers
REVIEW_REQUEST='{"type":"review-request","task_id":"'$TASK_ID'","deadline":"'$(date -u -d '+24 hours' +%Y-%m-%dT%H:%M:%SZ)'"}'
for REVIEWER in "${REVIEWERS[@]}"; do
pilotctl --json send-message "$REVIEWER" --data "$REVIEW_REQUEST"
done
Submit Review
REVIEW='{"type":"review-submission","task_id":"'$TASK_ID'","decision":"approve","score":0.92}'
pilotctl --json send-message "$REQUESTER_ADDR" --data "$REVIEW"
Finalize Review
APPROVALS=$(echo "$REVIEWS" | jq -r '[.[] | select(.data.decision == "approve")] | length')
[ $APPROVALS -ge $MIN_APPROVALS ] && echo "APPROVED" || echo "REJECTED"
Workflow Example
#!/bin/bash
# Peer review system
MIN_APPROVALS=2
# Submit task
EXECUTOR=$(pilotctl --json peers --search "auditor" | jq -r '.[0].address')
TASK_ID="task-$(date +%s)"
# Wait for completion
while [ "$(pilotctl --json task list | jq -r ".[] | select(.task_id == \"$TASK_ID\") | .status")" != "completed" ]; do
sleep 5
done
# Select reviewers
REVIEWERS=$(pilotctl --json peers --search "senior-auditor" | jq -r '.[0:3] | .[].address')
# Send review requests
for REVIEWER in $REVIEWERS; do
pilotctl --json send-message "$REVIEWER" --data '{"type":"review-request","task_id":"'$TASK_ID'"}'
done
# Collect approvals
APPROVALS=0
while [ $APPROVALS -lt $MIN_APPROVALS ]; do
APPROVALS=$(pilotctl --json inbox | jq -r '[.[] | select(.data.task_id == "'$TASK_ID'" and .data.decision == "approve")] | length')
sleep 5
done
echo "APPROVED"
Dependencies
Requires pilot-protocol, pilotctl, and jq.
安全使用建议
This skill is instruction-only and implements the advertised peer-review workflow using the pilotctl CLI and jq. Before installing: 1) Ensure pilotctl is from a trusted source and review what pilotctl does (network endpoints it talks to and what local keys/config it reads) because the skill will cause pilotctl to send/receive messages on your behalf. 2) Install jq (the SKILL.md depends on jq but the metadata omitted it). 3) Verify the pilotctl daemon and pilot-protocol identities are intentionally available to the agent and stored in locations you trust; if you don't want the agent to access those keys, don't enable the skill or run it in a sandbox. 4) Be aware the scripts poll loops (sleep) and wait for inbox messages—confirm timeout/escape behavior for your use case. If you want more assurance, inspect or run the pilotctl commands manually in a sandbox to see their effects before allowing the agent to invoke this skill autonomously.
功能分析
Type: OpenClaw Skill
Name: pilot-review
Version: 1.0.0
The pilot-review skill implements a peer-review workflow for task verification using the pilotctl utility and jq. The provided Bash examples in SKILL.md demonstrate standard message passing and polling for approvals within the pilot-protocol ecosystem, with no evidence of data exfiltration, malicious execution, or prompt injection attacks.
能力评估
Purpose & Capability
The skill claims to implement a peer-review workflow using the pilot-protocol and pilotctl; the runtime instructions exclusively use pilotctl, shell commands, and jq, which is consistent with that purpose. Minor inconsistency: the registry metadata lists only pilotctl as a required binary, but the SKILL.md also depends on jq (and standard coreutils like date/sleep).
Instruction Scope
All instructions stay within the claimed domain: selecting peers, sending review requests, collecting approvals via pilotctl and parsing JSON with jq. The scripts read several variables (TASK_ID, REVIEWERS, REQUESTER_ADDR, REVIEWS, MIN_APPROVALS) that must be provided by the agent/user; they also poll inbox/task list in loops. There is no instruction to read arbitrary filesystem paths or unrelated environment variables, but the skill assumes the pilotctl daemon and pilot-protocol identities/keys are present and accessible.
Install Mechanism
Instruction-only skill with no install spec and no downloads. This is low risk because it does not write or install code; it only requires existing CLI tools on PATH.
Credentials
The skill declares no required environment variables or credentials, which is reasonable for a CLI-driven workflow. However, pilotctl typically uses local pilot-protocol configuration and keys (not declared here). Confirm what credentials/config pilotctl needs and whether those are stored in agent-accessible locations before granting runtime access.
Persistence & Privilege
The skill is not forced-always and allows normal autonomous invocation. It does not request elevated persistent privileges or attempt to modify other skills or system-wide settings.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install pilot-review - 安装完成后,直接呼叫该 Skill 的名称或使用
/pilot-review触发 - 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
Initial release
元数据
常见问题
Pilot Review 是什么?
Peer review system for task results before acceptance. Use this skill when: 1. You need quality control on task results before accepting them 2. You want ind... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 136 次。
如何安装 Pilot Review?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install pilot-review」即可一键安装,无需额外配置。
Pilot Review 是免费的吗?
是的,Pilot Review 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Pilot Review 支持哪些平台?
Pilot Review 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 Pilot Review?
由 Calin Teodor(@teoslayer)开发并维护,当前版本 v1.0.0。
推荐 Skills