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/vietanhdev/bulwark/agents-mdgit clone --depth 1 https://github.com/vietanhdev/bulwarkWhat 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 | $0.13637 | $0.13637 |
| Opus 5 | $0.06818 | $0.06818 |
| Sonnet 5 | $0.02727 | $0.02727 |
| Haiku 4.5 | $0.01364 | $0.01364 |
Grade B, and why
bulwark AGENTS.md scanned grade B with 2 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 2d 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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
Two different mechanisms, deliberately: the GUI uses `pkexec` with `polkit/com.bulwark.policy` (`auth_admin`, one prompt per privileged scan — `auth_admin_keep` was dropped because it caches authorization for the *generi Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
**Secret-pack validation (new, and it found three real bugs).** The pack had 262 vendored regexes and no test that any individual rule still *worked* — `every_bundled_rule_compiles` proves a pattern parses, not that it c How it starts
The opening of the file, as written. The whole thing — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
This file provides guidance to AI coding agents working with code in this repository.
What this is
Bulwark scans a Linux host for security misconfigurations and intrusion indicators using a native Rust rule engine over declarative YAML rules, and explains findings in plain language with a suggested fix. Design rationale, architecture, and alternatives-considered all live in docs/guide/architecture.md — read that before making an architectural change, not just this file. Background research grounding the rule checklist (Lynis, MITRE ATT&CK, HackTricks) is in research/2026-07-11-linux-security-checklist/report.md.
Build & development commands
# Core + CLI
cargo build --workspace
cargo test --workspace
cargo clippy --workspace --all-targets -- -D warnings
cargo fmt --all # cargo fmt --all -- --check in CI
cargo run -p bulwarkctl -- scan
cargo run -p bulwarkctl -- rules validate rules/
cargo test -p bulwarkctl --test e2e -- --ignored --test-threads=1 # needs Docker; see below
# GUI (from apps/bulwark-app/)
npm install
cargo tauri dev # hot reload for frontend; Rust/tauri.conf.json changes need a restart
cargo tauri build # produces .deb, .rpm, .AppImage in target/release/bundle/
npx tsc --noEmit # type-check the frontend
npm run lint # eslint
npm run format:check # prettier --check
# Docs site (from docs/)
npm install
npm run dev # local preview
npm run build # static build to docs/.vitepress/dist
# Packaging (from workspace root, after a release build)
cargo build --release -p bulwarkctl
cargo deb -p bulwarkctl --no-build # requires `cargo install cargo-deb`
cargo generate-rpm -p crates/bulwarkctl # requires `cargo install cargo-generate-rpm`
CI (.github/workflows/ci.yml) runs fmt-check, clippy -D warnings, cargo test --workspace, rules validate rules/, and a frontend typecheck — run all of these locally before considering a change done.
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.
- 2d ago First seen · 239 lines · 13,637 tokens per session scan B d05f386cb3c8
bulwark AGENTS.md is an instructions file published in the GitHub repository vietanhdev/bulwark (11 stars, last pushed 9d ago), licensed Apache-2.0. It adds 13,637 tokens to every session, about $0.0682 per session on Opus 5. A static security scan graded it B with 2 findings (asks for root, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
lite AGENTS.md
AGENTS.md instructions for ultralytics/lite, covering agents.md, core principles (critical), pr workflow, commands and architecture.
lite CLAUDE.md
Claude Code instructions for ultralytics/lite, a project described as: A fast, local workspace for Claude Code, Codex, Gemini CLI, Kimi Code, Qwen Code, DeepSeek, OpenRouter, and your shell.
noid-privacy-workstation AGENTS.md
AGENTS.md instructions for NexusOne23/noid-privacy-workstation, covering noid privacy workstation — cross-agent engineering policy, 1. policy role, ownership, and distribution, 2. decision model: boundaries, user intent, and autonomy, 3. authority, evidence, and uncertainty and 4. product scope, licensing, and…
moonproxy-desktop AGENTS.md
Instructions for MoonProxyHQ/moonproxy-desktop, covering moonproxy, 功能, 技术栈, 目录结构 and 配置与数据存储.
mcp-screenshot-server CLAUDE.md
Instructions for sethbang/mcp-screenshot-server, covering claude.md, what this is, commands, environment variables and architecture.
Token-Dashboard CLAUDE.md
Instructions for Arylmera/Token-Dashboard, covering claude.md, project overview, status, architecture and data source.