paper-explain-figures

Original：🇨🇳 Chinese

Translated

2 scripts

Interpret the meaning of paper figures and output a highly readable Markdown report that 'teaches humans how to read figures'; supports input of absolute paths to one or more figure files and manual interpretations, automatically attempts to retrieve the source code used to generate the figures from the vicinity of the figures, and uses a parallel-vibe-like approach to interpret each figure with process-level isolation via `codex exec`/`claude -p` (default concurrency limit is 3, adjustable in config.yaml). ⚠️ Not applicable: Users only want to adjust figure size/crop/change format; or request direct modification of images/source code (this skill has read-only access to images and source code throughout, modification is strictly prohibited).

4installs

Sourcehuangwb8/chineseresearchlatex

Added on2026-05-05

NPX Install

npx skill4agent add huangwb8/chineseresearchlatex paper-explain-figures

SKILL.md Content (Chinese)

View Translation Comparison →

paper-explain-figures

Collaboration Agreement with bensz-collect-bugs

For bugs caused by design flaws of this skill, first use
```
bensz-collect-bugs
```
to record them in
```
~/.bensz-skills/bugs/
```
in a standardized manner, do not directly modify the skill source code installed locally by the user; if there is a workaround, record the bug first, then continue to complete the task.
Only when the user explicitly requests actions like "report bensz skills bugs" for public submission, use the local
```
gh
```
to upload new bugs to
```
huangwb8/bensz-bugs
```
; do not pull / clone the entire repository.

Objectives

For one or more paper figures provided by the user:

Convert the figures to
```
.jpg
```
(for visual understanding; if conversion fails, retain the original image and explicitly state this)
Automatically retrieve the "source code used to generate the figure" near the figure (may not be found; set to
```
NULL
```
if not found)
Output a highly readable report integrating three types of information:
- Understanding 1: Visual understanding (directly read the figure from jpg/original image)
- Understanding 2: Source code understanding (based on code; code determines the true meaning of the figure)
- Understanding 3: User's manual interpretation (helps guess the user's focus; may be incorrect)

Strict Constraints (Must Be Followed)

Hard Constraint on Directory Management: All intermediate products must be hosted in the hidden directory .paper-explain-figures/
under the current working directory (this directory name is hard-coded in the script; redirecting to other locations is not allowed).
Hard Constraint on Runtime Isolation: Runtime auxiliary files such as HOME, TMP, XDG cache/state/config generated by the runner / image converter must also be redirected and stored in
```
.paper-explain-figures/
```
.
Read-Only Constraint: Only read-only access is allowed to the user's figures and source code files throughout the process; modifying them (including formatting/rewriting/overwriting) is strictly prohibited.

Input

The user will input one or more pieces of information:

Absolute paths to one or more figure files
Optional: Manual interpretation of the figures (global or per-figure)
Optional: The user explicitly provides the absolute path to the source code file (prioritized as a "candidate entry")

Output

Final report: By default, output to
```
paper-explain-figures_report.md
```
in the current working directory
All intermediate files and logs: Stored in
```
.paper-explain-figures/
```
(saved in directories by run/job)

Standard Report Format (Must Follow This Structure for Each Figure)

Each figure is separated by a

##

level heading in the same Markdown file:

markdown

# Figures

## Figure: xxx

> File location: xxx
> Source code: xx.R lines xxx-xxx

### Core Meaning of the Chart

...

### Variable Definitions

| Element | Definition |
| --- | --- |
| ... | ... |

### Key Interpretation Points

1. ...
2. ...

### Explanation

...

### Scientific Value

...

Technical Approach (Process-Level Isolation + Concurrency Limit)

To ensure "mutual independence of interpretation for each figure", this skill uses worker scripts to launch independent processes in the shell for execution:

```
codex exec "..."
```
(Recommended: Easier for local file reading + visual understanding)
```
claude -p "..."
```
(Optional)

The default concurrency limit is 3 (adjustable in

paper-explain-figures/config.yaml

).

Usage Recommendations

Run in the current directory (recommended, products will be stored in the current directory and

.paper-explain-figures/

):

bash

python3 paper-explain-figures/scripts/paper_explain_figures.py \
  --fig /abs/path/to/figure1.png \
  --fig /abs/path/to/figure2.pdf \
  --note "Your focus/background explanation for these figures (optional)"

If the skill has been installed system-wide (path varies by platform, the following is only an example):

bash

python3 ~/.codex/skills/paper-explain-figures/scripts/paper_explain_figures.py --fig /abs/path/to/figure.png

Common parameters (as needed):

bash

# Enable parallelism (serial by default to reduce API rate limiting/banning risks)
python3 paper-explain-figures/scripts/paper_explain_figures.py --fig /abs/path/to/figure.png --parallel

# Concurrency limit (default 3; can also modify config.yaml:defaults.max_parallel)
python3 paper-explain-figures/scripts/paper_explain_figures.py --fig /abs/path/to/figure.png --parallel --max-parallel 3

# Specify runner (codex/claude/local)
python3 paper-explain-figures/scripts/paper_explain_figures.py --fig /abs/path/to/figure.png --runner codex

⚠️ Security Note:

--runner shell

has been disabled because it cannot provide strict guarantees that "no intermediate files will be leaked outside .paper-explain-figures".

Cleanup Method

Execute in the trigger directory:

bash

rm -rf .paper-explain-figures

paper-explain-figures

NPX Install

Tags

SKILL.md Content (Chinese)

paper-explain-figures

Collaboration Agreement with bensz-collect-bugs

Objectives

Strict Constraints (Must Be Followed)

Input

Output

Standard Report Format (Must Follow This Structure for Each Figure)

Technical Approach (Process-Level Isolation + Concurrency Limit)

Usage Recommendations

Cleanup Method