What is Agent-Security-Regression-Harness AGENTS.md?

A Python command-line tool that runs security checks against AI agents and systems connected through MCP, a standard for linking tools to AI applications.

Use it to run and validate YAML security scenarios, test HTTP or Python targets, record agent activity, and evaluate results such as tool calls and messages.

It checks whether security rules still hold after code or prompt changes, helping catch regressions before they reach users.

Agent-Security-Regression-Harness AGENTS.md — Instructions file for Codex and OpenCode by OWASP: A Python command-line tool that runs securi

Install

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.

agentmods

npx agentmods add instructions/owasp/agent-security-regression-harness/agents-md

Clone the repo

git clone --depth 1 https://github.com/OWASP/Agent-Security-Regression-Harness

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/owasp/agent-security-regression-harness/agents-md.svg)](https://agentmods.dev/instructions/owasp/agent-security-regression-harness/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/owasp/agent-security-regression-harness/agents-md"><img src="https://agentmods.dev/badge/instructions/owasp/agent-security-regression-harness/agents-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 927 This file is loaded in full into every session.

When invoked 927 The same file — it is already loaded in full.

Security scan A 0 findings. Scan, not verified.

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	$0.00927	$0.00927
Opus 5	$0.00464	$0.00464
Sonnet 5	$0.00185	$0.00185
Haiku 4.5	$0.00093	$0.00093

Measured 4d ago against content hash 143f39a0d184, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

Agent-Security-Regression-Harness 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 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.

AGENTS.md · 78 lines

How it starts

The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AI Agent Guide: OWASP Agent Security Regression Harness

This file helps AI coding agents (Claude, Copilot, Cursor, etc.) understand this project and contribute effectively.

Project overview

A Python CLI tool (agent-harness) that runs executable security regression scenarios against AI agents and MCP-integrated systems. It checks that security policies hold after code/prompt changes.

pip install -e . && agent-harness run scenarios/goal_hijack/basic.yaml --dry-run

Architecture

src/agent_harness/
  cli.py          # Entry point. argparse-based. Subcommands: version, validate, run, suite
  scenario.py     # Loads & validates YAML scenarios (Scenario dataclass)
  trace.py        # Trace dataclass (messages, tool_calls, events)
  assertions.py   # Evaluates assertions against traces. Each assertion = one function
  result.py       # HarnessResult + AssertionResult dataclasses, status aggregation
  recorder.py     # TraceRecorder helper for incremental trace building
  runner.py       # Orchestrates: load scenario -> run target/adapter -> eval assertions
  adapters.py     # HTTP target + Python callable target runners
  openai_agents_adapter.py  # Runs OpenAI Agents SDK Agent, converts result -> Trace
  langchain_adapter.py      # Runs LangChain/LangGraph invoke(), converts result -> Trace
  mcp_adapter.py            # Runs MCP workflow callable, converts result -> Trace
  mcp_runtime.py            # MCP server config validation (future full host)

tests/
  test_*.py       # pytest. Mirrors src structure.

Contribution workflow

Pick an open issue with help wanted or good first issue label
Fork, create a focused branch (feature/, fix/, scenario/, docs/)
Make small, reviewable changes
Run python -m pytest before pushing
Open PR and disclose AI assistance per CONTRIBUTING.md

What needs doing

Area	How to help
Scenarios	Add YAML files under `scenarios/<category>/`. See `docs/scenario-spec.md`
Assertions	Add function in `assertions.py`, register in `evaluate_assertions()`
Adapters	Add new target type in `adapters.py` or a new dedicated `*_adapter.py`
CLI	Add flags in `cli.py::build_parser()` + wiring in `main()`
Tests	Add `test_*.py` in `tests/` using pytest
Docs	Improve `docs/`, README, examples
CI	Add GitHub Actions workflows (`.github/workflows/`)

Read the full file on GitHub · 78 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.

4d ago First seen · 78 lines · 927 tokens per session scan A 143f39a0d184

Subscribe to this mod's changes

Agent-Security-Regression-Harness AGENTS.md is an instructions file published in the GitHub repository OWASP/Agent-Security-Regression-Harness (49 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 927 tokens to every session, about $0.0046 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.

Security Testing agent-security ai-security appsec llm-security mcp owasp

Agent-Security-Regression-Harness AGENTS.md

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

AI Agent Guide: OWASP Agent Security Regression Harness

Project overview

Architecture

Contribution workflow

What needs doing

What this file has done since we first saw it

Other instructions, from other repositories

arcjet-js AGENTS.md

promptsonar AGENTS.md

llm-safe-haven CLAUDE.md

cordon AGENTS.md

cyvisguard AGENTS.md

appsec-advisor AGENTS.md