What is hybrid-rag-mcp AGENTS.md?

An AGENTS.md instruction file for a Python MCP server that searches documents using keyword and optional meaning-based search. It also describes testing, evaluation, protection against prompt injection, and tracing.

Starting, testing, evaluating, and safely modifying the hybrid-rag-mcp project.

It gives coding agents the project's purpose, commands, file map, and rules that must not be broken before they edit the code.

How udarshmarthala/hybrid-rag-mcp configures Codex and OpenCode

Reuse

Borrowing it

Nothing to install: this file belongs to udarshmarthala/hybrid-rag-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/udarshmarthala/hybrid-rag-mcp/master/AGENTS.md

Clone the repo

git clone --depth 1 https://github.com/udarshmarthala/hybrid-rag-mcp

Made for: Codex, OpenCode.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/udarshmarthala/hybrid-rag-mcp/agents-md/github.svg)](https://agentmods.dev/instructions/udarshmarthala/hybrid-rag-mcp/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/udarshmarthala/hybrid-rag-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/udarshmarthala/hybrid-rag-mcp/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/udarshmarthala/hybrid-rag-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/udarshmarthala/hybrid-rag-mcp/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 2,538 This file is loaded in full into every session.

When invoked 2,538 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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 5.1	$0.02538	$0.02538
Opus 5	$0.01269	$0.01269
Sonnet 5	$0.00508	$0.00508
Haiku 4.5	$0.00254	$0.00254

Measured 8d ago against content hash fc038c9442bb, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

hybrid-rag-mcp AGENTS.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 8d 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.

AGENTS.md · 171 lines

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.

AGENTS.md

Machine-facing context for hybrid-rag-mcp. If you are an AI/coding agent, read this file first — it is the dense single-source entrypoint. Human docs live in docs/; this file summarizes them and adds the invariants you must not violate.

What this project is

A production-minded Model Context Protocol (MCP) server exposing one tool, search_docs, backed by hybrid retrieval (BM25 + optional dense vectors fused with Reciprocal Rank Fusion) and optional cross-encoder reranking — with an evaluation harness (recall@k, MRR), an input prompt-injection guardrail (OWASP LLM01), and OpenTelemetry tracing. Language: Python 3.10+. Transport: stdio.

Run / verify (fast path)

python3.13 -m venv .venv && source .venv/bin/activate   # needs Python 3.10+
pip install -e ".[dev]"        # core (BM25) + dev tools;  ".[full,dev]" adds dense+rerank+otel
make test                      # pytest -q            (unit tests)
make eval                      # retrieval recall@k + MRR table
make redteam                   # guardrail catch-rate report
make run                       # start MCP server over stdio (python -m hybrid_rag_mcp.server)

Smoke-test the tool without a client:

from hybrid_rag_mcp.server import search_docs
search_docs("how many approvals to deploy to prod", top_k=3)  # -> list[dict]

Invariants — DO NOT BREAK

mcp is pinned >=1.6,<2.0. The server uses the high-level FastMCP API (from mcp.server.fastmcp import FastMCP). MCP 2.0 removed FastMCP in favor of MCPServer. Do not upgrade past 2.0 without rewriting server.py to the new API.
Doc and Hit (in retrieval/__init__.py) are the retriever contract. Every retriever returns list[Hit]; fusion/rerank/eval depend on these shapes. Keep them stable.
Graceful degradation is mandatory. Heavy deps (sentence-transformers, flashrank, opentelemetry) are optional. Core install must work with BM25 only. Guard any heavy import with try/except + an is_available() function; construct the component only when available.
The guardrail runs first. RetrievalPipeline.search calls scan(query) before any retrieval and raises ValueError if flagged. Do not move retrieval ahead of it.
Stable identifiers. Guardrail rule names (_RULES) and span names (search_docs, retrieve.bm25, retrieve.dense, rerank) appear in reports/traces — renaming them is a breaking change.
Fusion is rank-based (RRF). Do not add score normalization to combine BM25 and dense; RRF fuses by rank on purpose. RRF only runs when >1 ranking exists.

Read the full file on GitHub · 171 lines

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.

8d ago First seen · 171 lines · 2,538 tokens per session scan A fc038c9442bb

Subscribe to this mod's changes

hybrid-rag-mcp AGENTS.md is an instructions file published in the GitHub repository udarshmarthala/hybrid-rag-mcp (0 stars, last pushed 29d ago), licensed MIT. It adds 2,538 tokens to every session, about $0.0127 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 2026-08-31.

Data and AI RAG and vectors

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

hybrid-rag-mcp: Instructions file for Codex