Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/shabaraba/vibing.nvim/nvim-contextnpx skills add shabaraba/vibing.nvim --skill nvim-contextgit clone --depth 1 https://github.com/shabaraba/vibing.nvimWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00075 | $0.01058 |
| Opus 5 | $0.00037 | $0.00529 |
| Sonnet 5 | $0.00015 | $0.00212 |
| Haiku 4.5 | $0.00007 | $0.00106 |
Grade A, and why
nvim-context scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Neovim Live Context
When the vibing-nvim MCP server is available, a real Neovim instance is running and its
in-memory state (open buffers, splits, cursor position, unsaved edits) is the ground truth —
it can differ from what's on disk. Prefer live state over assumptions whenever the user
references "this file", "current buffer", "my selection", "what I have open", etc.
Calling the tools
Two things decide whether a vibing-nvim tool call reaches the editor the user is looking at.
Both apply to every skill and subagent in this plugin, so they are stated once here.
Which name. Tools are written below as mcp__vibing-nvim__<tool>, which is the plain
user-level MCP server registration. Loaded as a Claude Code plugin — which is how vibing.nvim
itself provides them, handing the CLI this directory with --plugin-dir — they appear as
mcp__plugin_vibing-nvim_vibing-nvim__<tool> instead. If the plain prefix is not available, look
for a tool whose name ends in the one you need rather than assuming it is missing.
Which instance. Every tool takes an rpc_port naming the target Neovim.
- Inside a vibing.nvim chat, the port is in your system prompt for the turn — pass that exact value on every call.
- A subagent does not inherit it. Take it from your task prompt, and if it isn't there, call
nvim_list_instancesand use the port it reports. - Anywhere else (an ordinary Claude Code session that loaded this plugin some other way — a
--plugin-dirof your own, or a leftover install),nvim_list_instancesis the only way to know. If it lists more than one, say which you found and ask rather than guessing.
Omitting rpc_port works only while exactly one Neovim is live: reads fall back to the instance
registry, and writes refuse outright. Worktrees and concurrent chats make more than one the normal
case, so treat the fallback as a diagnostic, not a default.
Workflow
- Ground yourself first. Call
mcp__vibing-nvim__nvim_get_infofor the active file andmcp__vibing-nvim__nvim_list_windows/mcp__vibing-nvim__nvim_list_buffersto see everything open across splits/tabs before deciding which file the user means. - Use the real selection. If the user mentions a visual selection, call
mcp__vibing-nvim__nvim_get_visual_selectioninstead of asking them to paste code. - Respect unsaved state.
mcp__vibing-nvim__nvim_get_bufferreturns the buffer's current content, which may include unsaved edits that differ from the file on disk — read the buffer, not the file, when a buffer for that path is already loaded. - Cursor-relative requests. For "here", "at my cursor", "this function" type requests, use
mcp__vibing-nvim__nvim_get_cursorto resolve the exact line/column before acting. - Actual edits still go through your normal file tools (Read/Edit/Write). The vibing-nvim MCP tools are for observing and controlling the live editor (buffers, windows, commands), not a substitute for making code changes.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 68 lines · 75 tokens per session scan A 3448f6b192c9
nvim-context is a skill published in the GitHub repository shabaraba/vibing.nvim (13 stars, last pushed 3d ago), licensed MIT. It adds 75 tokens to every session and 1,058 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
add-sponsor
Add, move or remove a sponsor logo in the README and on the install page. Use when a GitHub sponsor reaches the Team or Company tier, when a sponsorship lapses, or when a sponsor sends a new logo asset.
code-review
Pre-commit code review using code-reviewer agent for bug detection, security analysis, and quality assurance. Manual invocation only. Use before git commits to catch issues early. Blocks commits on REQUESTCHANGES verdict.
debug
Systematic debugging using error-detective and debugger agents instead of manual investigation. Triggers automatically on: error, bug, broken, failing, exception, stack trace, test failure. Orchestrates RCA-first approach with parallel agent investigation.
deep-context
Build deep codebase understanding using Capsule context, progressive-reader, and specialist agents instead of overwhelming main context. Triggers on: don't have context, understand codebase, learn about, need background. Implements progressive context building.
workflow
Systematic task orchestration for complex multi-step tasks. Triggers automatically when detecting: complex task, multi-step work, coordinate, orchestrate, break down. Guides through Understand → Strategy → Plan → Execute → Verify phases for comprehensive systematic approach.
task-router
Decision matrix for choosing optimal approach to any task. Helps Claude decide when to delegate to sub-agents vs. working directly. Use when starting a new task or unsure of best approach.