kami-lel/kaye-library
A ready-to-install library of reusable LLM instructions: shared rules for writing, git, repositories, and coding, authored once in Markdown and ready to use as skills, plugins, or plain prompts, so any agent works to the same high standard from its first prompt, powered by kaye-engine
Defines the **Suite** and its **Movement** divisions — a multi-round workflow and the reviewable units it's broken into
Defines the **Movement** and its inner **Count** — reviewable division of large work, atomic action measuring progress — with naming, sizing, boundaries, and a\b🧩c\d prefix
Generates Chicago Manual of Style bibliography entries: source-template citation lines and annotated entries for a Works Cited or reference list
Generates Chicago Manual of Style notes-bibliography footnotes: kebab-case slugged markers, source-template citations, annotated notes, collected as native Markdown footnotes at file end
Gathers every source cited in a document and builds its Chicago-style bibliography list.
coordinates citation and bibliography assembly across the other Amanuensis skills
Runs all remaining Movements in 1 Turn, stops only when the Suite is done or the user must step in
Sets how you conduct a turn-based conversation: reply language, discourse terms, decorative emojis, and how a question is marked
Handles coding tasks — expansion, adjustment, debugging, and support — following strict formatting, naming, and comment conventions.
Handles coding tasks — expansion, adjustment, debugging, and support — following strict formatting, naming, and comment conventions.
writes Bash script files, functions, and sourced libraries for Debian GNU/Linux under a fixed house style.
answers Bash command requests for Debian GNU/Linux with the bare terminal line
Writes, edits, and reviews C code under a fixed house style: C99 and brace placement.
Writes, edits, and reviews C# code under a fixed house style: brace placement and XML documentation comments.
writes command usage messages — the compact call-shape reference a command prints for --help, in errors, or atop a manual page
writes comment banners inside source code — ranked heading-plus-ruler separators marking file titles, top-level sections, and nested subsections behind each language's comment leader
writes banners into printed program output — ranked heading-plus-ruler dividers separating run titles, execution phases, and nested checks across logs, demos, and test runs
Writes, edits, and reviews C++ code under a fixed house style.
Writes, edits, and reviews GDScript for Godot 4 under a fixed house style.
Use this skill when writing or generating HTML — apply HTML5 standards for structure, semantics, and markup. Trigger for any task that produces or edits .html files or embedded HTML content.
Writes, edits, and reviews all JavaScript and TypeScript code, targeting the ES11 standard with camelCase naming and JSDoc documentation conventions.
Applies a fixed PowerShell code style to PowerShell being written, edited, or reviewed.
writes, edits, reviews Python code: scripts, modules, packages, functions, classes, inline snippets
writes, formats Python docstrings in Sphinx/reStructuredText
writes, reviews Python pytest test code per project conventions
Writes, edits, and reviews Unreal Engine C++ under a fixed house style.
Commits the indexed diff with a primary summary and a per-file ledger.
Writes the message of a merge in progress: a typed first line prepended to Git's own message.
Shrinks an existing AGENTS and CONTEXT pair: reads both whole, drops finished plans, dead workflows, superseded detail, anything the code already answers, tightens the rest, adds nothing
Condenses CHANGELOG.md's [Unreleased] section for release: rewords its entries into concise, high-level changelog style and adds Admonitions for Breaks and Hazards where warranted.
Writes & maintains structured, hierarchical personal notes, specializing in capturing an app's, game's or system's preferences, settings & config state as a reproducible snapshot.
Bootstraps a CHANGELOG.md for a repo that has none — reconstructing dated, versioned sections from git history newest-first, with an [Unreleased] block and version links.
Creates AGENTS.md for a generic project that has none
Creates topic-focused docs/ files for a generic project
Use this skill when the user wants to create a new README.md from scratch — covering project overview, setup, usage, configuration, and contributing guidelines. Trigger even for casual requests like "write a readme" or "document this project."
Creates AGENTS.md for a repository that has none
Creates topic-focused docs/ files for a repository
Writes & maintains structured, hierarchical personal notes under 1 consistent filename, framing & prose grammar.
Drafts and processes emails and messages on the user's behalf.
Circled letters (ⓐ, ⓑ, ⓒ, ~) as a choice/subset label
Formats a date carrying no year, day of week prefix optional
Formats a time of day on a 24-hour clock, precision from minute down to sub-second
Defines the Watch Day, a day kept on a 30-hour clock that runs past midnight rather than ending at it
Defines the Watch Week, a forward planning range of days closing on a terminal Sunday
Formats a date carrying a year, day of week prefix optional
Provides accurate IPA transcription in /slash notation/ inline after any word requiring pronunciation clarity, across all languages.
Presents numerical values in both metric and US units with correct unit abbreviations and space-separated thousands.
Defines triage tags (TT) — defect/note labels spanning code and docs across 3 case tiers (Loud/Steady/Quiet), with per-tag meanings and raise/lower tier shifts
Equips project for AI agents: writes & maintains its AGENTS.md
Executes approved Plan file, 1 Movement per turn — Movement In, work, present with link back, Movement Out. Plan file doubles as saved state, so pause and resume carry across sessions
Decodes common programming abbreviations such as ctor, iface, and mpl.
Maps the single-letter day-of-week codes used in dates and schedules to their weekday names.
Decodes language codes such as en, zh, and jp to the natural languages they name.
Maps short language codes such as py, ts, and cpp to the programming languages they name.
Names the four Eisenhower matrix quadrants by Important and Urgent, with their emoji shorthand.
Decodes unit abbreviations such as ft, lb, and mi to the units they name.
Runs the user's daily morning check in one pass, closing with a single combined report.
Runs the user's weekly look ahead, events and job opportunities alike, closing with a single report.
Stops after 1 Movement, waits for user to open next Turn
Gathers job opportunity information from any available source and formats it into briefs, one per job, each with a title-and-company heading, key details, and a briefness-style overview.
Bootstraps a repository's entire baseline documentation set in one pass — README.md, CHANGELOG.md, AGENTS.md, CONTEXT.md, and docs/ — from a single repository inspection, splitting content by purpose with no duplication and skipping files that already exist.
Records an app's, game's or system's keybindings as a reproducible note in compact key symbols.
Lands a feature branch on dev: runs every Ready for Feature Landing movement, then merges the branch into dev with a typed Feature Landing message.
Use this skill when the user wants to add, fix, or reorganize entries in an existing CHANGELOG — logging new features, bug fixes, or breaking changes without overwriting existing content. Trigger even for casual requests like "update the changelog" or "log what changed."
Updates an existing project AGENTS.md against what changed, editing in place
Updates existing project docs/ files against what changed, editing in place
Updates an existing README to match the current project state — correcting stale overviews, commands, links, configuration names, and examples, pruning obsolete or duplicated content, and editing in place against the latest repository changes.
Updates an existing repository AGENTS.md against what changed, editing in place
Updates existing repository docs/ files against what changed, editing in place
Logs a session's durable changes before context compaction — recording them in CHANGELOG.md and syncing AGENTS.md and CONTEXT.md — identifying the change set once from the conversation, working diffs, and existing [Unreleased] entries.
Writes & maintains structured, hierarchical meeting minutes, specializing in tracking a meeting's decisions & action items as a reproducible, followed-through record.
Splits oversized task into ordered, numbered Movements, then runs them under review — survey repository, write Plan file, hold for approval, execute 1 Movement per turn
Creates a task's linked worktree as REPO-WT-FEATURE in the repository's WS- workspace folder, on the task's feature branch, creating the branch if missing.
Writes & maintains AGENTS.md for generic AI-ready projects: rules & constraints telling agents how to work on project
Writes & maintains docs/ topic files for generic projects
Defines the semantic versioning scheme: how MAJOR, MINOR, and PATCH numbers are chosen and written.
Extends semantic versioning with conventions for alpha/beta/rc pre-release numbering, OS/date build-metadata tags, and version-to-dev-stage mapping (toy, vertical slice, pre-alpha, alpha, beta, rc, stable)
Writes a version release title from new vs. old version: bump word, release type, version.
Deletes local branches already merged into the current branch, with a report of what went and what stayed.
Retrieves existing events from the user's outside calendar app or service into the conversation, so the conversation can work against the user's actual schedule.
Retrieves existing tasks from user's GTD app into conversation, to work against actual to-do list.
Adds an event to the user's outside calendar app or service, from a single supplied event or from an event list already organized in the conversation.
Adds tasks to user's GTD app, each assigned a project & an Eisenhower quadrant, from 1 supplied task or a task list organized in conversation.
Records a feature branch's changes into CHANGELOG.md and updates AGENTS.md and CONTEXT.md to match, syncing all affected docs in one pass before a merge, then commits the work.
Cuts a project release: brings README.md, CHANGELOG.md, AGENTS.md, CONTEXT.md, and docs/ current, moves [Unreleased] changelog entries into a dated versioned section, and bumps the version across metadata files like package.json, pyproject.toml, Cargo.toml, or setup.cfg.
Recaps what user has finished today so far as one warm, encouraging Recap Line: events attended, mail answered, tasks closed, gathered from record & reported to cheer the user toward doing more.
Recaps what user finished yesterday as one warm, encouraging Recap Line: events attended, mail answered, tasks closed, gathered from record & reported to cheer the user toward doing more.
Revises user-provided text while preserving original intent and style, offering suggestions and iterating on feedback.
Releases the project: runs every Ready for Version Release movement, merges the branch into main with a typed release message, and tags the merge with its v version.
Writes & maintains repository AGENTS.md & personal AGENTS.local.md: **prescriptive** rules telling AI coding agents how to work in repo, e.g. commands, code style, PR rules, safety constraints
Writes and maintains CHANGELOG files per Keep a Changelog conventions — dated version entries newest-first, grouped change types, a persistent [Unreleased] section, and linkable version references.
CONTEXT.md is the **descriptive** knowledge layer: a dual-audience briefing for new developers and AI assistants on *what the codebase is* — architecture, data flow, domain model, API surface, patterns, conventions, known gaps. It goes deeper than the setup-focused README.md and is the foundational map other files build on; CONTEXT.local.md holds personal, gitignored context. This skill writes and maintains both.
Writes & maintains docs/ topic files for repositories
Writes and maintains human-friendly README files — scannable, visually clear landing pages covering a project's purpose, features, setup, usage, and contribution flow, linking every reference document inline within the explanation that names it, with a standard title format and tasteful use of headings, lists, badges, and emoji.
Defines a standard, language-agnostic project/repository layout — naming conventions and placement for top-level documentation files and source, build, docs, test, and tooling folders.
Resolves conflicts in a halted Git merge — reconciles both sides of every hunk into correct working-tree files, markers removed.
Audits whole repository for gaps, drift, unfinished seams — anything that becomes user-visible bug on public release; marks every surviving gap with a Loud TT.
Reworks running Plan file when reality stops matching it: halts, rewrites Plan, holds for approval before work resumes
Scans the event radar for today and formats what it finds into one daily brief.
Scans the event radar across the coming week and formats what it finds into one weekly brief.
Creates detailed academic reading notes from provided text — summarizing paragraph by paragraph into structured bullet points — and generates Chicago-style citations and bibliographies on request.
Formats a set of events into ## 📡 Event Radar briefs: an Event Heading, emoji-prefixed Key Details, and a briefness-style Event Overview per event.
Reorders existing triage tags into grouped blocks — tier from Loud to Quiet, then meaning from bug to fixme, todo, hack. Position only, nothing else changes.
Rewrites content in "Briefness Style" — terse, newspaper-headline prose that maximizes brevity: dropped articles and helper verbs, strong nouns and verbs, active voice, numerals and abbreviations, punctuation-compressed phrasing, no terminal periods.
Proofreads and polishes text with minimal edits — fixing spelling, grammar, punctuation, and clarity while preserving the original meaning, voice, and wording.
Sets blank-line spacing before Markdown headers, scaled by header level and file-length tier
Sets wide blank-line spacing before Markdown headers, for long files
Sets moderate blank-line spacing before Markdown headers, for medium-length files
Names the All Caps convention: uppercase words joined by underscores.
Names the Camel Case convention: first word lowercase, later words capitalized, no separators.
Names the Commentary Case convention: lowercase opening, standard capitalization after, no final punctuation.
Names the Kebab Case convention: lowercase words joined by hyphens.
Names the Pascal Case convention: each word capitalized, no separators.
Names the Snake Case convention: lowercase words joined by underscores.
Names the Title Case convention: major words capitalized, minor words lowercase.
Formats in GitHub Flavored Markdown: bold for key info, italics for titles/secondary info, no underscores
Attaches a footnote-style identifier link inline and collects every definition in an endnotes block at the file's end.
Formats dash-bulleted lists in commentary case, with optional Title-Case colon labels (bold sparingly, never italic); renders math in LaTeX (inline $…$, block $$…$$); renders diagrams/flowcharts in Mermaid code blocks
Attaches a Markdown footnote to each link, keeping the footnote's definition directly beside its reference.
Opens a pull request landing a feature branch on dev, after readying its docs, with the title and body drawn from the branch summary.
Sweeps an email inbox: reads every unread message, marks it read, sorts it into a category
Reconciles a failing test suite with a codebase's changed behavior — realigning stale assertions, fixtures, and baselines to the new contract, and surfacing the failures where the code, not the test, may be what drifted.
Places a feature branch's linked worktree, one per task, as REPO-WT-FEATURE inside the repository's WS- workspace folder, and removes it once its edits reach the main worktree.
Sets baseline for all Unity 6 C# work, the shared root every topic-specific Unity skill builds on
Places Unity assets & folders across Assets/, so any file is found by asset type, domain & module without search
Writes, edits & reviews Unity C# logic in house idiom, so code reads consistently & a debug session traces as a sequence of events
Assigns each Unity class, script, GameObject, or asset one architectural role, so responsibilities split cleanly & its name says what it does
Moves Unity gameplay numbers out of code into designer-editable data, so balance changes need no recompile
Declares Unity Inspector fields so designers read, tune & rename them without breaking saved values
Structures Unity MonoBehaviour scripts in one fixed layout, so every component reads alike & missing Inspector assignments surface at startup
Names anything in a Unity project, code identifier or asset, so related things sort together & each name reads at a glance
Controls Unity GameObjects living only part of a run through one owner, so showing, hiding & spawning stay predictable & cost nothing while idle
Turns Unity scene objects into prefabs & variant families early, so one edit updates every copy
Structures Unity ScriptableObject classes in one fixed layout, keeping asset data read-only & valid at runtime
Removes a finished task's linked worktree once every edit in it is committed and merged into the main worktree's branch.
Sets the Git working terms and the actions that need explicit approval, so edits stay in the workspace beside the user.
Pairs every Perforce term with its closest Git equivalent in parentheses, so a user who thinks in Git never translates alone.
Drafts and improves LLM system-message prompts, advises on prompt engineering, and preserves source syntax like JSON schema.
Writes a skill's description field: a self-contained statement of what the skill does, leading with the main verb and outcome.
Writes both a skill's description and when_to_use fields together, each per its own convention.
Writes a skill's when_to_use field: trigger phrases, keywords, and a boundary note disambiguating it from neighboring skills.