momo-land/PHD-management-skills / skills/phd-manager
phd-manager
Verified在 PHD-management-skills 中管理科研工作台,直接响应当前 AI 对话里的“一键更新”“一键导入”“更新论文进度”“更新实验进度”“生成周报”等短口令。读取当前对话与本地记录,执行相应技能并同步网页,也支持初始化和梳理研究下一步。
| File | Kind | Size |
|---|---|---|
| SKILL.en.md | markdown | 8.0 KB |
| SKILL.md | skill md | 7.2 KB |
| agents/openai.yaml | config | 290 B |
| references/batch-workflow.en.md | markdown | 3.4 KB |
| references/batch-workflow.md | markdown | 3.0 KB |
| references/local-workflow.en.md | markdown | 7.8 KB |
| references/local-workflow.md | markdown | 6.9 KB |
| references/research-workflows.en.md | markdown | 2.7 KB |
| references/research-workflows.md | markdown | 2.4 KB |
| references/upstream-skills.en.md | markdown | 1.6 KB |
| references/upstream-skills.md | markdown | 1.5 KB |
references/batch-workflow.en.md
View on GitHub# Batch writes for progress updates and material imports
Follow the [local workflow](local-workflow.en.md). The current AI prepares a batch from the user's authorization and actual material; no model API or client-specific command syntax is required.
## Protocol
```text
node cli.mjs batch --file <json-file> --dry-run
node cli.mjs batch --file <json-file>
```
`--file -` accepts standard input. The top-level object contains `operations`, with 1–100 sequential operations:
```json
{
"operations": [
{
"action": "create",
"collection": "tasks",
"data": {
"id": "new-task-id",
"projectId": "existing-project-id",
"title": "Task the user explicitly asked to record",
"status": "todo",
"origin": "user"
}
},
{
"action": "update",
"collection": "papers",
"id": "existing-paper-id",
"data": { "version": 3, "status": "revising" }
}
]
}
```
Replace example IDs and versions with actual values. List existing records first, omit unchanged operations, consolidate repeated changes to one record into one update, and preserve notes, association arrays and untouched fields. Do not reimport an existing record as a create.
- Supported actions are `create` and `update`, not deletion. A create may omit `data.id`; when a later operation must reference it, generate a valid, unused stable ID and include it explicitly.
- References can point to existing records or records created earlier in the same batch, not forward references. Put parents before dependents. For a paper and experiment requiring links in both directions, create the paper, create the experiment, then update the new paper's association using version 1.
- Update `id` is an operation-level field; `data.version` is the version just read. A new record starts at version 1. Versions are not numbers to increase arbitrarily to bypass conflicts.
- All database changes in one batch, including activity logs, succeed together or roll back together. Manuscript and other file operations are outside this transaction. More than 100 candidates may be split by dependencies into separate batches, but those batches are not one transaction. Back up first, verify each batch and report the exact completed scope; do not claim an atomic import of all material.
## A dry run is not a save
A successful preview returns `{ok:true,dryRun:true,applied:0,results:[...]}`. Application returns `dryRun:false` and the actual `applied` count. Results contain each operation's index/action/collection/record. A preview rolls back every write, and generated IDs or timestamps are not retained. Do not reuse preview-assigned random IDs for later operations or claim that a preview saved anything.
When the user has asked to update or import progress, apply the clear, authorized changes after validation without requesting duplicate approval. Application checks versions again. After a conflict, read and assess the new state rather than treating the successful preview as unconditional permission to overwrite. Errors identify the failed item; no items in that batch are saved.
After success, list/read state to verify changes, sources and associations. If readback fails, state the application result and the verification limitation; do not blindly retry creates before checking what was saved. An open webpage refreshes the shared state automatically, while the CLI also works without a running server.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |