auto-paper-improvement-loop
Autonomously improve a generated paper via GPT-5.4 xhigh review → implement fixes → recompile, for 2 rounds. Use when user says \"改论文\", \"improve paper\", \"论文润色循环\", \"auto improve\", or wants to iteratively polish a generated paper.
90
88%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Passed
No known issues
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong skill description that clearly communicates a specific multi-step workflow (review → fix → recompile × 2 rounds), includes both Chinese and English trigger terms, and has an explicit 'Use when' clause. The description is concise yet comprehensive, and the unique combination of paper polishing with specific tooling makes it highly distinctive.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists specific concrete actions: 'GPT-5.4 xhigh review → implement fixes → recompile, for 2 rounds.' This describes a clear multi-step pipeline with specific tools and iteration count. | 3 / 3 |
Completeness | Clearly answers both 'what' (autonomously improve paper via review → fix → recompile for 2 rounds) and 'when' (explicit 'Use when' clause with specific trigger phrases and a general condition). | 3 / 3 |
Trigger Term Quality | Includes natural trigger terms in both Chinese and English: '改论文', 'improve paper', '论文润色循环', 'auto improve', and the natural phrase 'iteratively polish a generated paper'. Good coverage of how users would actually phrase this request. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche: autonomous iterative paper improvement using a specific review model (GPT-5.4 xhigh) with recompilation. The Chinese trigger terms and specific workflow make it very unlikely to conflict with other skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, highly actionable skill for an iterative paper improvement loop with strong workflow clarity including validation checkpoints, state persistence for crash recovery, and concrete fix patterns. Its main weaknesses are moderate verbosity (some explanatory context and the typical score progression section could be trimmed) and the monolithic structure that could benefit from splitting reference material (fix patterns, prompt templates) into separate files.
Suggestions
Trim the Context section comparing to /auto-review-loop and the 'Typical Score Progression' section — these are informational rather than instructional and consume tokens without guiding action.
Consider extracting the fix pattern tables and review prompt templates into separate reference files (e.g., FIX_PATTERNS.md, REVIEW_PROMPT.md) and linking to them from the main skill.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is quite long (~300 lines) with some unnecessary verbosity — the Context section explaining differences from /auto-review-loop, the 'Typical Score Progression' anecdote, and explanatory comments could be trimmed. However, most content is actionable tables and concrete steps, so it's not egregiously padded. | 2 / 3 |
Actionability | Highly actionable with executable bash commands, specific prompt templates for the reviewer agent, concrete fix pattern tables mapping issues to solutions, and copy-paste ready compilation commands. The spawn_agent/send_input patterns give exact message formats. | 3 / 3 |
Workflow Clarity | Excellent multi-step workflow with clear sequencing (Steps 0-9), explicit validation checkpoints (verify 0 undefined references/citations after each compile, format check in Step 8 with auto-fix patterns), state persistence for recovery, and a feedback loop structure (review → fix → recompile → re-review). Human checkpoint option adds another validation layer. | 3 / 3 |
Progressive Disclosure | The content is entirely self-contained in one large file with no references to external documentation. The fix pattern tables and review prompt templates could be split into separate reference files. However, the use of collapsible <details> blocks in the log template and clear section headers provide reasonable internal organization. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Commit
dc00dfb
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.