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 instructions/sachinshelke/codevira/claude-mdgit clone --depth 1 https://github.com/sachinshelke/codeviraWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/sachinshelke/codevira/claude-md)<a href="https://agentmods.dev/instructions/sachinshelke/codevira/claude-md"><img src="https://agentmods.dev/badge/instructions/sachinshelke/codevira/claude-md.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.02807 | $0.02807 |
| Opus 5 | $0.01404 | $0.01404 |
| Sonnet 5 | $0.00561 | $0.00561 |
| Haiku 4.5 | $0.00281 | $0.00281 |
Grade A, and why
codevira CLAUDE.md 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 4d 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 — 171 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codevira — persistent project memory
This project uses Codevira to give every AI coding tool you use shared memory of the project. Decisions, fix history, the structural code graph, and learned style preferences persist across sessions and across IDEs.
When to call which Codevira tool
Call these MCP tools at the moments the description matches your action — they are fast (tens of ms), token-efficient (summary-by-default), and what makes the project's memory available to you.
At the start of every session
get_session_context()— Always call this first. Returns a ~500-token brief: current focus, recent decisions, top open items. Without it you're blind to the project's history.
Before modifying any file
-
get_impact(file_path)— Returns who calls this code (callers, dependents, blast radius). Required before any rename, signature change, or cross-cutting refactor. If callers > a small number, propose a deprecation+migration plan instead of a direct edit. -
search_decisions(query)— Returns prior architectural decisions touching the topic. Decisions withdo_not_revert: trueMUST be respected; if your proposed change conflicts, surface the decision to the user and wait for approval before proceeding.
Before adopting a pattern, library, or naming convention
-
get_session_context()— itsstylepanel carries the top communication preferences (~30 tokens), so match them without a separate call. For code style, read the surrounding code: naming, error handling and test layout are visible there and do not need a tool.(
search_preferenceswas removed in 4.0 — see "Removed in 4.0" below.)
When you fix a bug or land an architectural choice
record_decision(...)— Capture the why: what alternatives were considered, why this won, what would force a re-examination. A one-line decision is enough; the value compounds across sessions and across the other AI tools the user opens this project in.
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.
- 4d ago First seen · 171 lines · 2,807 tokens per session scan A 3c340b311c8a
codevira CLAUDE.md is an instructions file published in the GitHub repository sachinshelke/codevira (12 stars, last pushed 19d ago), licensed MIT. It adds 2,807 tokens to every session, about $0.0140 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 instructions, from other repositories
agents-memory AGENTS.md
AGENTS.md instructions for Lolaplex/agents-memory, covering agent install (this repo), procedure and done when.
ai-mind-map copilot-instructions.md
Copilot instructions for shdra06/ai-mind-map, covering ai mind map mcp — code memory engine (v1.4.0), 🚀 first thing to do in every conversation, quick lookup: "i need to..." → use this tool, ⚡ session lifecycle (always use these) and 🔍 find code (instead of grep/reading files).
vibe-memory CLAUDE.md
Claude Code instructions for shahrukh-hack/vibe-memory, covering claude code integration guide and memory lifecycle commands.
vibe-memory AGENTS.md
AGENTS.md instructions for shahrukh-hack/vibe-memory: This repository adheres to the Universal Agent Memory Protocol (AGENTMEMORY.md). All AI agents (Antigravity, Cursor, Claude Code, OpenAI Codex) share the same persistent memory context.
agents-memory CLAUDE.md
Claude Code instructions for Lolaplex/agents-memory, a project described as: Local markdown memory & cross-agent context engine for AI coding assistants (MCP + injection).
ai-mind-map CLAUDE.md
Claude Code instructions for shdra06/ai-mind-map, covering ai mind map mcp — code memory engine (v1.4.0), 🚀 first thing to do in every conversation, quick lookup: "i need to..." → use this tool, ⚡ session lifecycle (always use these) and 🔍 find code (instead of grep/reading files).