文件操作
文件系统与批量文件任务
agent-customization
microsoft
'WORKFLOW SKILL — Create, update, review, fix, or debug VS Code agent customization files (.instructions.md, .prompt.md, .agent.md, SKILL.md, copilot-instructions.md, AGENTS.md). USE FOR: saving coding preferences; troubleshooting why instructions/skills/agents are ignored or not invoked; configuring applyTo patterns; defining tool restrictions; creating custom agent modes or specialized workflows; packaging domain knowledge; fixing YAML frontmatter syntax. DO NOT USE FOR: general coding questions (use default agent); runtime debugging or error diagnosis; MCP server configuration (use MCP docs directly); VS Code extension development. INVOKES: file system tools (read/write customization files), ask-questions tool (interview user for requirements), subagents for codebase exploration. FOR SINGLE OPERATIONS: For quick YAML frontmatter fixes or creating a single file from a known pattern, edit the file directly — no skill needed.'
minimax-xlsx
MiniMax-AI
"Open, create, read, analyze, edit, or validate Excel/spreadsheet files (.xlsx, .xlsm, .csv, .tsv). Use when the user asks to create, build, modify, analyze, read, validate, or format any Excel spreadsheet, financial model, pivot table, or tabular data file. Covers: creating new xlsx from scratch, reading and analyzing existing files, editing existing xlsx with zero format loss, formula recalculation and validation, and applying professional financial formatting standards. Triggers on 'spreadsheet', 'Excel', '.xlsx', '.csv', 'pivot table', 'financial model', 'formula', or any request to produce tabular data in Excel format."
deft-directive-probe
deftai
Adversarial plan stress-testing via relentless one-question-per-turn interrogation. Use when the operator asks to run probe, /deft:run:probe, or wants a plan challenged before committing to it.
imaging-data-commons
ImagingDataCommons
Query and download public cancer imaging data from NCI Imaging Data Commons using idc-index. Invoke for any question about IDC collections, cancer imaging datasets, DICOM data access, radiology (CT, MR, PET) or pathology AI training sets, metadata queries, visualization, or license checks — even when the user doesn't explicitly mention "IDC". No authentication required.
metainformant-src-metainformant-protein-database
docxology
METAINFORMANT rules for directory src/metainformant/protein/database. Use when editing, adding tests, or reviewing code under this path. Read the linked AGENTS.md first; use uv only, write outputs to output/, real implementations.
agent-memory
rolandcanyon-cmd
Teach cross-session memory patterns using MEMORY.md — what to save, how to organize it, how to maintain it over time, and how to structure topic files as memory grows. Works in any Claude Code project with no external dependencies. Trigger words: remember this, save for later, across sessions, persistent memory, don't forget, note this, keep this, write this down.
room-edit
rolandcanyon-cmd
Edit a previously-catalogued room — rename, add/remove aliases, move devices between rooms, fix typos, mark devices inoperable, remove dead entries. Same review-before-commit pattern as /room-walk.
iconfont-downloader
JS-mark
This skill should be used when the user asks to "下载图标", "搜索图标", "iconfont", "download icon", "search icon", "找个图标", or when the user needs to search and download SVG icons from iconfont.cn.
agent-to-agent-bridge
joecastelino
Talk directly to other AMG agents (Walter II, Ralph, Don, etc.) from Jay's session via a CLI bridge. Use when Joe asks Jay to coordinate with, relay to, or ask another agent something on the same machine.
jay-google-sheets-drive-access
joecastelino
Access Joe's Google Sheets and Drive files from Jay's profile — token symlink bootstrap, reading/writing native Sheets, downloading Drive-hosted .xlsx files (like AMG WIP.xlsx) via alt=media, and the xlsx-vs-native-Sheet write limitation. Established 2026-08-02.
jay-gmail-draft-verification
joecastelino
Independently verify a Gmail draft/send that Stacey (email-agent) reports she created — confirm it's actually in Drafts/Sent with correct recipients, body, and attachments, without trusting her self-report alone. Use whenever Joe asks to "chase" or double-check an email Jay routed through Stacey, or when Jay's own Gmail-dependent flow (himalaya/Google API) throws auth errors.
jay-brain-and-skill-index
joecastelino
Operate, diagnose, and repair Jay's GBrain (the single Jay-only knowledge brain) plus the self-maintaining skill index. Use when the nightly brain refresh fails or looks frozen, when gbrain is "command not found", when GBrain search returns nothing/stale, when session knowledge needs to be captured into the brain so it's retrievable later, or when the skill index needs descriptions/metadata/usage-counts. Covers the bun-PATH break, the multi-brain split, the embedding-key gotcha, and the capture-session / skill-usage tracking scripts.
skill-optimizer
IonDen
Use when asked to shrink, trim, compress or optimize an agent skill (a folder with a SKILL.md) so it costs fewer tokens, or when a skill feels too long. Cuts by deletion and restructuring only, can move rarely needed sections into references/ behind a load trigger, and checks that every rule, exception, command and threshold survives before anything is applied. Never edits the frontmatter; suggests description changes instead. Subagent definitions in .claude/agents/ and Codex agents in .codex/agents/ are a different thing, handled by subagent-optimizer. Triggers: "optimize this skill", "shrink my skill", "this skill is too long", "trim SKILL.md", "my skills eat my context", "skill optimizer".
sd-grill
platypeeps
Use when the user wants their own proposal, plan, claim, or half-formed idea interrogated one question at a time until it survives an outsider, producing a hardened statement of intent, or an honest account of where it stopped, rather than an implementation.
sd-profile
platypeeps
Use when the user wants to create, inspect, correct, review, import, export, or forget a consent-driven personal operating profile with traceable assertions.
sd-coherence-audit
platypeeps
Use when a knowledge corpus — a note vault, agent-instruction files, or a docs tree — must be audited against itself for contradictions, vagueness, bandaid guidance, and redundancy, returning a read-only findings ledger with both sides quoted.
sd-propose-skills
platypeeps
Use when the user wants the current session reviewed for recurring friction, repeated steps, and hard-won gotchas, and high-bar skill proposals drafted into the report for the user to decide.
hmasd-peer-collaboration
CartmanFatass
Coordination between the Claude DM session and the Codex Root as peer researchers - channels, message shape, inbox convention, shared-working-tree hygiene and disagreement routing under constitution section 2. A method, not a role or an authority.
distributed-job-safety
terrylica
Concurrency safety patterns for distributed pueue + mise + systemd-run job pipelines.
session-recovery
terrylica
Diagnose and recover Claude Code session issues including missing conversations, corrupted sessions, and session file problems. Use whenever the.
disk-hygiene
terrylica
macOS disk cleanup, cache pruning, stale file detection, and Downloads triage. TRIGGERS - disk space, cleanup, disk usage
hook-authoring
Mike-Byrne-AI
Hook script authoring conventions for tools/cc/hooks/. Use when writing or extending a hook, adding a bash extraction pattern to write_guard, modifying _normalize_path or path-handling helpers, reasoning about hook event ordering, or when the user mentions hook authoring, hook contract, hook regression, or path traversal in the harness. Codifies the contract every hook script must satisfy.
project-directory-migration
terrylica
Migrate Claude Code project sessions when renaming directories. TRIGGERS - directory rename, move project, migrate sessions
hmasd-peer-collaboration
CartmanFatass
Coordination between the Claude DM session and the Codex Root as peer researchers - channels, message shape, inbox convention, shared-working-tree hygiene and disagreement routing under constitution section 2. A method, not a role or an authority.