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 rules/adonai-labs/agent-runway/engineering-securitygit clone --depth 1 https://github.com/adonai-labs/agent-runwayWhat 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 | $0.01032 | $0.01032 |
| Opus 5 | $0.00516 | $0.00516 |
| Sonnet 5 | $0.00206 | $0.00206 |
| Haiku 4.5 | $0.00103 | $0.00103 |
Grade A, and why
engineering-security 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 — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Engineering Security
Universal security principles applied whenever writing or reviewing code that handles input, identity, secrets, or sensitive data.
These principles apply regardless of language, framework, or technology stack.
Validate all input at the boundary
- Never trust caller-supplied data; validate type, format, range, and length before use
- Reject invalid input explicitly; do not attempt to sanitise and proceed
- Validation belongs at the entry point of the system, before any processing
Least privilege
- Code, services, and users should request only the permissions they need
- Default to restrictive; require explicit grant for elevated access
- If a service does not need write access to a database table, it should not have it
Authenticate before authorising
- Always verify identity before checking permissions
- Do not allow partial or anonymous access to sensitive operations unless intentional and documented
- Enforce authentication and authorisation at the boundary; do not rely on callers doing it correctly
Authorise at the resource level
- Check that the authenticated user is permitted to access the specific resource being requested
- Do not rely only on role checks; verify resource ownership in business logic
- Missing authorisation on one endpoint is sufficient for an access control failure
Secrets belong in secrets management
- Never store secrets in source code, configuration files committed to source control, or environment variables set in code
- Inject secrets from a secrets store at runtime (vault, environment, or managed identity)
- Rotate secrets on compromise; prefer short-lived tokens over long-lived credentials
Protect against injection
- Use parameterised queries or ORM methods for all database access; never concatenate user input into SQL
- Validate and escape output in rendering contexts; do not trust stored data as safe to render
- Treat all external input — headers, query parameters, request bodies, file names — as potentially hostile
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 · 137 lines · 1,032 tokens per session scan A 74f123ab095a
engineering-security is a cursor rule published in the GitHub repository adonai-labs/agent-runway (2 stars, last pushed 12d ago), licensed MIT. It adds 1,032 tokens to every session, about $0.0052 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 cursor rules, from other repositories
archgate-governance
Archgate ADR governance — enforces architecture decision records.
cursorrules
When the user asks you to generate diagrams, analyze code structure, or visualize architecture, follow this analysis pipeline. Use terminal commands (rg, find, cat) for all file operations.
arch4
Keep Arch4 architecture maps current.
cursorrules
You are working with a React Native (Expo) architecture reference project. Read the skill files before generating any code.
archcore-files
Enforce MCP-only operations when working with .archcore/ files.
execution
Repository execution and verification commands.