What does this skill do, and when should you use it?
This skill helps researchers standardize experiment logging by converting uploaded images, audio, or text into Markdown files with YAML frontmatter, while archiving raw attachments. It supports output to a local folder or an Obsidian vault, with optional Feishu CLI integration for pulling group messages. The core workflow requires no external services and keeps all data local. Ideal for labs that need consistent, searchable, and traceable experiment records.
Accepts user-uploaded images, audio, or text, or reads local file paths; extracts structured info via visual analysis and text parsing, confirming ambiguous fields with the user; generates experiment and sample batch IDs; writes Markdown logs to a specified output root and archives raw attachments to raw/experiments; optionally integrates Obsidian templates (index, anomaly log, equipment tracking) and Feishu CLI to fetch group messages; finally reports generated file and raw material locations.
- A grad student in a materials lab who wants to log daily corrosion immersion experiments with photos and notes in a structured way.
- An electrochemistry researcher who wants Dataview to auto-generate experiment lists and anomaly summaries in Obsidian.
- A team sharing experiment images and voice messages in a Feishu group, wanting automated aggregation into ID-tagged logs.
- Researchers who need to trace the same sample batch across multiple days, keeping sample_batch consistent for relational queries.
How do you install this skill?
- Publisher identity is unverified; review the skill content yourself before installation and use.
- Depends on unspecified model capabilities like vision_analyze; actual output quality may vary by model, recommend testing on real experiment materials.
- Optional Feishu integration requires bot permissions; ensure minimal bot permissions and be mindful of organizational data security.
- No rollback mechanism provided; if files are written locally, manual handling needed to undo, ensure backups.
- Shell / CLI
- Local filesystem
Install via npx skills: npx skills add Yuan1z0825/nature-skills --global --agent codex --skill nature-experiment-log --yes --copy, or clone the repo and manually copy the skills/nature-experiment-log directory to your agent's skill folder.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/Yuan1z0825/nature-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/nature-experiment-log" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Log today's corrosion immersion experiment; images uploaded; output to my Obsidian vault.
After installation, in an agent session, upload experiment images, audio, or text and specify the output method (e.g., Obsidian vault path or local folder). Example: "Log today's corrosion immersion experiment; images uploaded; output to my Obsidian vault." For Feishu integration, ensure the bot is added to the target group with proper permissions.
What are this skill's strengths and limitations?
- Structured YAML frontmatter enables programmatic search and Dataview queries
- Raw materials are auto-archived for traceability
- Flexible output to plain folders or Obsidian
- Explicitly asks for missing info instead of guessing, ensuring data integrity
- Marked as Draft; not yet tested on real cases
- Requires user-defined system and equipment codes, adding initial setup overhead
- Feishu integration depends on a separate skill and needs extra permission configuration
- No automated test suite; edge cases may be problematic
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Standardized Experiment Logging for Nature-style Research this page | 46 · Use with care | ★ 47k | 3d ago | Apache-2.0 |
| Trackio Experiment Tracking Skill | 53 · Use with care | ★ 38k | 1mo ago | MIT |
| II-Commons Retrieval Skill | 56 · Use with care | ★ 16 | 4mo ago | Apache-2.0 |
| Trackio Experiment Tracking | 43 · Not recommended | ★ 11k | 3d ago | Apache-2.0 |
| TAO DAFT Dataset Converter ✓ NVIDIA · Official | 54 · Use with care | ★ 3.5k | 3d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill explicitly requires user confirmation for output method and target directory, and returns save-ready Markdown without arbitrarily choosing a path when unspecified, showing some least-privilege and confirmation awareness. The core workflow does not depend on external services; optional Feishu integration needs bot permissions and is only used when needed, reducing overreach risk. However, there is no detailed disclosure of data-flow transparency (e.g., handling of sensitive experiment data) and no explicit rollback mechanism. Publisher identity is unverified, but the repository has an Apache-2.0 license and clear maintainer info, adding attribution clarity. Deductions because permission confirmation and rollback are incomplete.
Instructions are generally self-consistent with clear process steps, but rely on unspecified capabilities like vision_analyze, and lack tests or error-handling details. Static review cannot verify execution, so reliability cannot exceed 10; further deduct due to lack of explicit failure-feedback mechanisms and abnormal-input handling.
The skill has a clear target audience (experimental researchers) and provides rich trigger examples (upload images, audio, text), with good Chinese support suitable for Chinese users. However, capability boundaries are not precise, e.g., no explicit statement of unsupported input types or unhandleable cases. Environment fit is good, core function does not rely on overseas services, but Feishu and Obsidian integrations require extra configuration, potentially limiting some users. Deductions because boundaries and trigger conditions are not precise.
Documentation is well-structured with README, SKILL.md, manifest.yaml, references, examples, and templates, with reasonable information architecture. License is clear (Apache-2.0), version exists (manifest.yaml 1.0.2), but no changelog and explicit known-limitations disclosure. Maintenance responsibility is mentioned in README but the skill's own update path is unclear. Deductions because version history and known limitations are incomplete.
The skill design is reasonable and can accomplish the core task of standardizing experiment logs with a clear output format (Markdown+YAML). However, no actual execution evidence, static evaluation cannot verify direct usability. Examples are fictional, no real-case validation. Deductions because evidence of effectiveness and comparative advantage is lacking.
Only repository CI tests (for other skills) exist, no tests for this skill. Examples are all fictional, no third-party verification. Static review cannot independently reproduce, insufficient evidence, hence low score.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →