为Paperclip新增或扩展贯穿真实浏览器、服务器、数据库和runner的端到端评估,整理测试数据、判定条件与证据,并在现有报告中查看操作结果。
- 新增完整用户操作流程的评估
- 更新测试数据与判定条件
- 在评估报告中加入截图和产物
检查Paperclip的Mine收件箱,将已完成或长期无变化的条目列为整理候选,仅归档用户通过复选框确认的选项,并提供撤销所需的API操作。
Scan a Paperclip user's Mine inbox, classify reversible archive candidates, request checkbox confirmation, and archive only accepted selections. Use when asked to garden, clean up, prune, or tidy a Paperclip inbox without changing issues, branches, or workspaces.
检查Paperclip的Mine收件箱,整理并展示归档候选。附带脚本会将条目分为关联工作已完成、长期无变化、仍有领先于合入目标的提交,以及需要保留四类。只有用户通过复选框确认的条目才会归档,结果还会列出用于撤销归档的API操作。
适合清理仍留在收件箱中的已完成工作通知,或重新检查较旧的条目。仍有未合入提交的工作也会纳入检查,由用户自行选择要整理的内容。收件箱归档只改变当前用户的显示状态,不会修改原始任务。
需要Node.js、PAPERCLIP_API_URL和PAPERCLIP_API_KEY。按scan、confirm、apply的顺序执行,确认完成前不会归档。不会删除分支或工作区。如果搜索达到获取上限,会提示检查范围可能不完整。
本介绍由 AI(OpenAI Codex)根据公开的 SKILL.md 撰写。准确的规格请查看下方原文。
AI 仅根据 SKILL.md 的内容,从 4 个方面进行了评估。星标数和热度不计入评估。
总分按每项 A=25 分、B=15 分、C=5 分相加。 由 AI(OpenAI Codex)于 2026年10月20日 评估。仅供参考,安装前请同时查看原文。
安装前可以查看交给 AI 代理的指令内容。
Use the bundled script for every stage. Keep the workflow strictly ordered: scan → confirm → apply.
scan as read-only. It may read inbox, workspace, close-readiness, and local Git metadata only.confirm as a confirmation-card write only. It must not archive inbox entries or mutate issue fields, branches, or workspaces.apply only after a resolved request_checkbox_confirmation interaction. It archives only accepted option IDs that also occur in the originating candidates.json.Require PAPERCLIP_API_URL and PAPERCLIP_API_KEY. The script strips a trailing /api from the URL. It resolves the target user from the run JWT payload's responsible_user_id; use --user-id <uuid> only for an explicit target override.
Use a run-owned output directory when available:
RUN_DIR="${PAPERCLIP_RUN_SCRATCH_DIR:-${PAPERCLIP_TASK_SCRATCH_DIR:-.}}/garden-inbox"
mkdir -p "$RUN_DIR"
node .agents/skills/garden-inbox/scripts/garden-inbox.mjs scan \
--output-dir "$RUN_DIR" \
--stale-days 60
Inspect garden-inbox-report.md and candidates.json. The report groups every inbox row into exactly one bucket:
The scan reads Mine separately for each included issue status so the endpoint's global 500-row cap does not silently omit older rows. If any single-status query reaches that cap, both outputs mark coverage as possibly truncated; do not treat that scan as complete.
Do not manually promote bucket D entries into the candidate file.
Post checkbox interactions on the driving issue:
node .agents/skills/garden-inbox/scripts/garden-inbox.mjs confirm \
--issue-id "$PAPERCLIP_TASK_ID" \
--candidates "$RUN_DIR/candidates.json"
The script posts sequential cards when a scan has more than 200 candidates. Re-running confirm with the same scan file is idempotent. Leave the driving issue in the waiting posture required by the surrounding Paperclip heartbeat workflow.
When a candidate was declined by the user in an earlier pass, pass --unselect <issueId> (repeatable) so it starts unchecked and its description notes the earlier decline. Never re-offer previously declined items as default-checked.
For development or payload review, suppress the POST:
node .agents/skills/garden-inbox/scripts/garden-inbox.mjs confirm \
--issue-id "$PAPERCLIP_TASK_ID" \
--candidates "$RUN_DIR/candidates.json" \
--dry-run
After an interaction-resolution wake, take the resolved interaction ID from the wake payload and run:
node .agents/skills/garden-inbox/scripts/garden-inbox.mjs apply \
--issue-id "$PAPERCLIP_TASK_ID" \
--interaction-id "$INTERACTION_ID" \
--candidates "$RUN_DIR/candidates.json"
On rejection, expiry, or an accepted empty selection, the script archives nothing. Apply preserves the scan file's target user, including an explicit --user-id override. Its summary includes the API undo path and target-user body for every archived row.
Test apply without API writes by supplying a saved interaction response:
node .agents/skills/garden-inbox/scripts/garden-inbox.mjs apply \
--issue-id "$PAPERCLIP_TASK_ID" \
--interaction-file resolved-interaction.json \
--candidates "$RUN_DIR/candidates.json" \
--dry-run
Run the zero-dependency Node tests after changing classification or selection safety:
node --test .agents/skills/garden-inbox/scripts/garden-inbox.test.mjs
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
为Paperclip新增或扩展贯穿真实浏览器、服务器、数据库和runner的端到端评估,整理测试数据、判定条件与证据,并在现有报告中查看操作结果。
为Paperclip Runner添加或扩展行为评估,定义判定条件、执行目标和报告验证数据,利用正常及异常证据验证结果,并保留执行记录与费用。
通过实际浏览器操作网页和应用,检查界面显示与运行情况。结合截图、简单表单填写、错误及网络请求记录,帮助复现问题并确认操作结果。
通过 Paperclip 分配的 AgentMail 收件箱,查看任务关联邮件、发送新邮件或回复,并核查投递结果;重试时保留相同标识,避免重复发送。
检查GitHub、GitLab和Perforce变更中的评审意见、自动检查失败及说明缺漏,整理待办事项,并按请求协助修改和解决已处理的讨论。
根据构想或现有代码仓库,梳理AI智能体的职责、汇报关系与工作交接,生成符合规范的公司或团队文件包,供组织协作和导入Paperclip使用。