Agent Skills · Knowledge Management
Curated skills for Claude, Cursor and other AI agents. Copy one prompt to install.
Knowledge-management skills update notes, wikis, or docs so they match a change in code or in a decision. The good ones name a folder and a format. The bad ones offer to reorganize everything you have ever written.
A documentation updater aimed at one framework, such as a Next.js docs skill, belongs here when your repository is that framework and the docs live in a place you can diff. Point it at a branch. Ask it to document one change you just made, and see whether it edits only docs. Install and remove shows a trial built around exactly that test.
Do not confuse this shelf with memory plugins. A plugin stores state inside DeepSeek Harness. A skill on this shelf edits files in the project. If those files are committed, your teammates inherit the prose. That is a feature when the prose is right and a mess when the skill invented a feature flag.
Personal knowledge bases and company wikis raise an access question. A prompt that can "tidy the vault" can also merge two people's notes or paste a secret from one note into another. Run it on a copy. Secrets in a vault are still secrets.
Search in the directory is not knowledge management. A search hit is a card. This category is for skills that write or maintain documents. If you only needed to find a card, you are done. If you needed the docs in the repo to change, pick one skill and review the diff the way you would review a teammate's.
Commit the docs diff separately from the code diff so a reviewer can reject the prose without rejecting the fix. If the skill mixed them, split the commit yourself. Mixed commits are how a wrong sentence ships with a right patch.
Updates Next.js documentation based on code changes in the active branch to help maintainers review PR documentation completeness.
Turn a topic or outline into a structured Markdown mind map for common visualization tools.
Extracts an eight-layer reasoning fingerprint from scholar texts and applies a nine-module constraint to run peer review, supervision, lecturing, or panel discussion with quantitative scores and qualitative comments.
Guides users through a nine-step failure review to analyze causes, hidden gains, and controllable improvements, producing a structured Word report.
Paste a paper abstract or passage and output a structured note with subject, methods, innovations, conclusions, and limitations or outlook.
Fast CLI for summarizing URLs, local files, and YouTube links with multi-provider models and JSON output.
Open-ended visual QA for images, combining CV and LLM for scene description, chart interpretation, reasoning, and cloud history lookup.
Safely organizes loose files into extension-based folders, with restore and preview modes plus conflict-safe renaming and reports.
Automatically compiles Feishu documents with quality checks, deduplication, Wiki updates, and multidimensional table index writes.
A generic knowledge review skill that generates role-based checklists, rrule-based reminders, and on-demand reviews without storing personal data.
A knowledge-management skill for layered retrieval across information sources.
Turns Taleb's antifragility, barbell strategy, and via negativa into executable prompts for risk analysis in career, investing, health, and other domains.