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.
git clone --depth 1 https://github.com/stackhawk/agent-skillsWrote 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.
[](https://agentmods.dev/rules/stackhawk/agent-skills/stackhawk-hawkscan-input-vectors)<a href="https://agentmods.dev/rules/stackhawk/agent-skills/stackhawk-hawkscan-input-vectors"><img src="https://agentmods.dev/badge/rules/stackhawk/agent-skills/stackhawk-hawkscan-input-vectors/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.
<a href="https://agentmods.dev/rules/stackhawk/agent-skills/stackhawk-hawkscan-input-vectors"><img src="https://agentmods.dev/badge/rules/stackhawk/agent-skills/stackhawk-hawkscan-input-vectors.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00000 | $0.01425 |
| Opus 5.5 | $0.00000 | $0.00570 |
| Sonnet 5.5 | $0.00000 | $0.00285 |
| Haiku 4.5 | $0.00000 | $0.00143 |
Grade A, and why
stackhawk-hawkscan-input-vectors scanned grade A with 1 finding 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 1mo 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
**Recipe.** Record one real request per non-JSON operation — a curl against the running app, a How it starts
The opening of the file, as written. The whole thing — 109 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Input Vectors Reference
Active injection rules fire only on requests that carry an input and reach a live sink. A
spec that renders the wrong request body, a path parameter with no real id behind it, or a
GET-only crawl leaves whole rule families with nothing to test — and the scan reports nothing,
which reads as "clean". This file gives the three recipes that fix that. Field syntax is
canonical in hawk config show <field-path> --text; the YAML below shows shape only, with
placeholders you must replace.
Contents
- Rule: active injection needs a reachable sink
- HAR seed for XML and other non-JSON request bodies
- customVariables for per-resource values
Rule: active injection needs a reachable sink
An active rule (SQL injection, XSS, XXE, XPath, command injection) injects into a parameter and watches the response. That only works when:
- the request reaches the handler — a real id or parent resource exists (
GET /p/<id>with an id that is in the database, not a generated one), and - the request carries the parameter in the shape the handler parses — JSON for a JSON handler, XML for an XML handler, a query string for a query handler.
If either fails, the handler 404s or 500s before it parses input, and the rule reports
nothing. That is an unexercised vector, not a clean result. Seed data (the stackhawk-data-seed
skill, SKILL.md Phase 1c.6) fixes an empty database; the two recipes below fix the body shape
and the id when the spec cannot express them.
HAR seed for XML and other non-JSON request bodies
Why. The OpenAPI request builder emits a JSON body for operations declared as
application/xml. The XML handler 500s on it, so the XXE rule (and every other rule on that
operation) never runs. With a HAR that carries a real application/xml POST, the XXE rule runs.
Recipe. Record one real request per non-JSON operation — a curl against the running app, a
browser session, or hawk perch browser — and import the HAR alongside the spec. HAR
supplements openApiConf; it does not replace it.
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.
- 1mo ago First seen · 109 lines · 0 tokens per session scan A 70b935ba459a
stackhawk-hawkscan-input-vectors is a cursor rule published in the GitHub repository stackhawk/agent-skills (16 stars, last pushed 9d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,425 tokens. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-17.
Other cursor rules, from other repositories
api-property-optionality-hygiene
Fix ApiProperty/ApiPropertyOptional optionality mismatches in DTO files; use for scheduled batch fixes or DTO edits.
python
Python best practices and patterns for modern software development with Flask and SQLite.
cursor
You are working on the checkout service. Preserve transaction integrity and auditability.
django
Definitive guidelines for writing maintainable, performant, and secure Django applications, emphasizing modern best practices, clear code organization, and efficient patterns.
shared-libraries
Shared libraries - condition framework, inventory containers, file-backed DB, itinerary, references.
selectools-providers
Rules for working with LLM provider adapters in selectools.