Bruteforce sinks (adversarial-payload reachability)

--bruteforce-sinks statically drives a fixed set of hostile payloads at every risky call site and reports each payload that reaches the call without a recognised guard.

vulkro scan . --bruteforce-sinks
vulkro scan . --bruteforce-sinks --bruteforce-categories sql,payment,llm

Nothing is executed and no network call is made: it is a walk over your code. It is off by default because it is the slowest pass. In the local console, turn it on in the scan run options. Languages: Python, JavaScript and TypeScript.

What it adds to taint analysis

Taint analysis says an input reaches stripe.charges.create. This pass says which kinds of input get there, for example:

• amount = -1 reaching a payment call with no sign check
• ../../admin reaching s3.getObject with no path check
• ignore previous instructions reaching an LLM call with no system and user role split
• ' OR 1=1 -- reaching cursor.execute with no parameter binding

Sink categories

Tag	Examples
sql	cursor.execute, Sequelize.query, knex.raw, prisma.$queryRaw
shell	subprocess.run(shell=True), child_process.exec, os.system
http	requests.get, httpx.post, axios.*, fetch
payment	stripe.charges.create, braintree.transaction.sale, razorpay.orders.create
llm	anthropic.messages.create, openai.chat.completions.create, bedrock.invoke_model
file-write	open(path, 'w'), Path.write_text, fs.writeFile
deserialization	pickle.loads, yaml.load, eval, vm.runInNewContext
email	sendgrid.send, ses.send_email, smtplib.sendmail

Payload classes (--bruteforce-payload-classes): type-confusion, numeric-extreme, size-extreme, unicode, injection, path-traversal, prompt-injection, payment-extreme.

Recognised guards include type and null checks, length and range checks, anchored regexes, allowlists, sanitisers (html.escape, DOMPurify.sanitize, shlex.quote), parameter binding, path canonicalisation, schema validation (Pydantic, Zod, Joi, Yup, AJV) and yaml.SafeLoader.

Confidence

Level	Meaning
High (default)	No recognised guard on the path.
Medium	A guard exists but is not shown to cover this payload class. Opt in with --bruteforce-confidence medium.

Rule IDs

BFS-<CATEGORY>-<payload-tag>, for example BFS-SQL-sql-or-1-eq-1 or BFS-PAYMENT-payment-negative-amount. IDs are stable across runs, so baselines can key on them.

Environment variables

Variable	Default	Meaning
VULKRO_BRUTEFORCE_SINKS	unset	1 turns the pass on
VULKRO_BRUTEFORCE_DEPTH	6	Call depth, 2 to 12
VULKRO_BRUTEFORCE_PATHS_PER_FN	64	Paths per function, 8 to 512
VULKRO_BRUTEFORCE_CATEGORIES	all	Comma-separated tags
VULKRO_BRUTEFORCE_CONFIDENCE	high	high or medium
VULKRO_BRUTEFORCE_PAYLOAD_CLASSES	all	Comma-separated tags

See also: Attack paths, Taint analysis