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/zanzythebar/cursor-rules/agents-mdgit clone --depth 1 https://github.com/ZanzyTHEbar/cursor-rulesWhat 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 | $0.00881 | $0.00881 |
| Opus 5 | $0.00441 | $0.00441 |
| Sonnet 5 | $0.00176 | $0.00176 |
| Haiku 4.5 | $0.00088 | $0.00088 |
Grade A, and why
cursor-rules 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 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.
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 — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Repo Shape
- This is a Go CLI. Entry point:
cmd/cursor-rules/main.go. - Keep package boundaries intact:
internal/cliwires Cobra/Viper and rendering,internal/appowns use-case orchestration and destination/config resolution,internal/coredoes filesystem/install/watch work,internal/transformowns Cursor/Copilot conversions,internal/securityguards path handling. - Native package layouts are not symmetric:
- Rules come from root
*.mdcfiles or package directories. - Commands come from
commands/<name>/orcommands/<name>.md, butinstall commands ...writes Cursor-compatible skills into.cursor/skills/, not.cursor/commands/. - Skills come from
skills/<name>/SKILL.md. - Agents come from
agents/<name>.md; code still falls back to legacyagent/if present. - Hooks come from
hooks/<preset>/hooks.jsonplus scripts.
Commands
- Build/run:
make build,go run ./cmd/cursor-rules <subcommand>. - Focused test packages:
go test ./internal/cli/...,go test ./internal/core/...,go test ./cmd/cursor-rules/.... - CI-like verification:
go vet ./...,make test,make test-coverage,make build, plusgolangci-lint runif installed. - Install/sync/watch changes: run
scripts/integration_test.sh. - Root e2e workflow:
make test-e2e. - Do not use
make test-unit; it still targets./cli/...and currently fails because the code lives underinternal/cli.
Config And Runtime Gotchas
CURSOR_RULES_PACKAGE_DIRchanges both the source package dir and the default config-dir base: unlessCURSOR_RULES_CONFIG_DIRis set, config defaults to the parent of the package dir.- Destination precedence is
--dir user|global>--dir <path>>--global>--workdir. - Any CLI command can auto-start the watcher if config has
watch: true; this happens in rootPersistentPreRunE, not only incursor-rules watch. - Watcher auto-apply only writes presets listed in
watcher-mapping.yaml.autoApply: truewithout a mapping still watches but skips writes. sync --applyusesconfig.presetsif present; otherwise it applies every shared preset.install allwith default--target cursoralso installs commands, skills, agents, and hooks, not just rules.CURSOR_RULES_SYMLINK=1andCURSOR_RULES_USE_GNUSTOW=1change cursor-target install strategy;enableStow: trueonly takes effect whenstowis onPATH.cursor-rules config linkis the supported way to exposeCURSOR_*_DIRoverrides back under~/.cursor/....
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 · 40 lines · 881 tokens per session scan A a07c821435fb
cursor-rules AGENTS.md is an instructions file published in the GitHub repository ZanzyTHEbar/cursor-rules (3 stars, last pushed 4mo ago), licensed MIT. It adds 881 tokens to every session, about $0.0044 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.
Other instructions, from other repositories
kraken AGENTS.md
AGENTS.md instructions for luisabwk/kraken, covering agent instructions, the 3-layer architecture, why this works, your role and example flow.
kraken CLAUDE.md
Claude Code instructions for luisabwk/kraken, covering agent instructions, the 3-layer architecture, why this works, your role and common commands.
kleosrules AGENTS.md
Instructions for kleosr/kleosrules, a project described as: Cursor harness pack: user rules, skills, Bash hooks, local HANDOFF memory. macOS, Linux, Windows (WSL).
AI-Powered-Coding-Tools CLAUDE.md
Claude Code instructions for dereknguyen269/AI-Powered-Coding-Tools, covering claude.md, what this repo is, maintenance scripts, validate links in a markdown file and fix markdown formatting issues.
syncai copilot-instructions.md
Instructions for dmtrkzntsv/syncai, covering claude.md, build & run, architecture, lifecycle and identify is the routing decision.
tju-thesis-review-kit CLAUDE.md
Instructions for ChenXi-hub/tju-thesis-review-kit, covering claude.md, 工作原则, 审查优先级, 硬约束 and 可复用技能.