fix(tools): address review — close async bypass, fix false positives

- Sanitize subagent callback results in subagent.go (the primary
  attack vector from #972 — async results bypassed ExecuteWithContext)
- Remove generic hex pattern that matched git SHAs, Docker digests,
  and checksums in go.sum/package-lock.json
- Order regex patterns most-specific-first to prevent partial matches
- Add Stripe key patterns (sk_live_, sk_test_, rk_live_, rk_test_)
- Fix IsSecretEnvVar false positives on AUTHOR, AUTHORITY, etc.
  by switching from substring to suffix matching
- Preserve error chain via Unwrap() on redactedError
- Use strings.Contains instead of custom helpers in tests
- Add idempotency test, git SHA preservation test, checksum test
This commit is contained in:
Subash 2026-03-13 09:41:03 +05:30
parent 0026955946
commit 886d66911e
3 changed files with 96 additions and 57 deletions

View file

@ -6,16 +6,17 @@ import (
) )
// secretPatterns matches common API key and credential formats. // secretPatterns matches common API key and credential formats.
// Each pattern is compiled once at init and reused across calls. // Ordered from most specific to least specific to avoid partial matches.
// Each pattern is compiled once at package init and reused across calls.
var secretPatterns = []*regexp.Regexp{ var secretPatterns = []*regexp.Regexp{
// OpenAI / OpenAI-compatible (sk-...)
regexp.MustCompile(`sk-[A-Za-z0-9_-]{20,}`),
// OpenAI project keys (sk-proj-...)
regexp.MustCompile(`sk-proj-[A-Za-z0-9_-]{20,}`),
// Anthropic (sk-ant-...)
regexp.MustCompile(`sk-ant-[A-Za-z0-9_-]{20,}`),
// OpenRouter (sk-or-v1-...) // OpenRouter (sk-or-v1-...)
regexp.MustCompile(`sk-or-v1-[A-Za-z0-9_-]{20,}`), regexp.MustCompile(`sk-or-v1-[A-Za-z0-9_-]{20,}`),
// Anthropic (sk-ant-...)
regexp.MustCompile(`sk-ant-[A-Za-z0-9_-]{20,}`),
// OpenAI project keys (sk-proj-...)
regexp.MustCompile(`sk-proj-[A-Za-z0-9_-]{20,}`),
// OpenAI / OpenAI-compatible (sk-...)
regexp.MustCompile(`sk-[A-Za-z0-9_-]{20,}`),
// Google AI / Gemini (AIza...) // Google AI / Gemini (AIza...)
regexp.MustCompile(`AIza[A-Za-z0-9_-]{30,}`), regexp.MustCompile(`AIza[A-Za-z0-9_-]{30,}`),
// GitHub tokens (ghp_, gho_, ghu_, ghs_, ghr_) // GitHub tokens (ghp_, gho_, ghu_, ghs_, ghr_)
@ -24,14 +25,19 @@ var secretPatterns = []*regexp.Regexp{
regexp.MustCompile(`xox[bpsa]-[A-Za-z0-9-]{20,}`), regexp.MustCompile(`xox[bpsa]-[A-Za-z0-9-]{20,}`),
// Discord bot tokens (base64.base64.base64) // Discord bot tokens (base64.base64.base64)
regexp.MustCompile(`[MN][A-Za-z0-9]{23,}\.[A-Za-z0-9_-]{6}\.[A-Za-z0-9_-]{27,}`), regexp.MustCompile(`[MN][A-Za-z0-9]{23,}\.[A-Za-z0-9_-]{6}\.[A-Za-z0-9_-]{27,}`),
// Generic long bearer/api tokens (40+ hex or base64 chars) // Stripe keys (sk_live_, sk_test_, rk_live_, rk_test_)
regexp.MustCompile(`\b[A-Fa-f0-9]{40,}\b`), regexp.MustCompile(`[sr]k_(?:live|test)_[A-Za-z0-9]{20,}`),
// AWS access keys (AKIA...) // AWS access keys (AKIA...)
regexp.MustCompile(`AKIA[A-Z0-9]{16}`), regexp.MustCompile(`AKIA[A-Z0-9]{16}`),
// AWS secret keys (40 char base64-ish after known prefixes) // AWS secret keys (40 char base64-ish after known prefixes)
regexp.MustCompile(`(?i)aws[_\-]?secret[_\-]?access[_\-]?key["'\s:=]+[A-Za-z0-9/+=]{40}`), regexp.MustCompile(`(?i)aws[_\-]?secret[_\-]?access[_\-]?key["'\s:=]+[A-Za-z0-9/+=]{40}`),
} }
// NOTE: Generic hex patterns (e.g. [A-Fa-f0-9]{40,}) are intentionally excluded.
// They match git SHAs, Docker digests, checksums in go.sum/package-lock.json, etc.
// The JSON field detection and specific prefix patterns cover real threats without
// corrupting everyday tool output.
// jsonSecretFields matches JSON keys that typically hold secrets. // jsonSecretFields matches JSON keys that typically hold secrets.
// Captures the key and value so we can redact the value in-place. // Captures the key and value so we can redact the value in-place.
var jsonSecretFields = regexp.MustCompile( var jsonSecretFields = regexp.MustCompile(
@ -63,7 +69,7 @@ func redactToken(token string) string {
return "[REDACTED]" return "[REDACTED]"
} }
// Show first 4 chars for identification, redact the rest. // Show first 4 chars for identification, redact the rest.
return token[:4] + "..." + "[REDACTED]" return token[:4] + "...[REDACTED]"
} }
// SanitizeResult applies secret redaction to both ForLLM and ForUser fields // SanitizeResult applies secret redaction to both ForLLM and ForUser fields
@ -80,20 +86,20 @@ func SanitizeResult(r *ToolResult) *ToolResult {
if r.Err != nil { if r.Err != nil {
cleaned := RedactSecrets(r.Err.Error()) cleaned := RedactSecrets(r.Err.Error())
if cleaned != r.Err.Error() { if cleaned != r.Err.Error() {
r.Err = &redactedError{msg: cleaned} r.Err = &redactedError{msg: cleaned, cause: r.Err}
} }
} }
return r return r
} }
// redactedError wraps a redacted error message. // redactedError wraps a redacted error message while preserving the error chain.
type redactedError struct { type redactedError struct {
msg string msg string
cause error
} }
func (e *redactedError) Error() string { func (e *redactedError) Error() string { return e.msg }
return e.msg func (e *redactedError) Unwrap() error { return e.cause }
}
// ContainsSecret checks whether a string contains any known secret pattern. // ContainsSecret checks whether a string contains any known secret pattern.
// Useful for pre-flight validation before logging or outputting content. // Useful for pre-flight validation before logging or outputting content.
@ -109,19 +115,20 @@ func ContainsSecret(s string) bool {
return false return false
} }
// envSecretKeys lists environment variable name patterns whose values should // envSecretKeys lists environment variable name suffixes whose values should
// never appear in tool output. Used by the shell tool guard. // never appear in tool output.
var envSecretKeys = []string{ var envSecretKeys = []string{
"API_KEY", "SECRET", "TOKEN", "PASSWORD", "CREDENTIAL", "_API_KEY", "_SECRET", "_TOKEN", "_PASSWORD", "_CREDENTIAL",
"ACCESS_KEY", "PRIVATE_KEY", "AUTH", "_ACCESS_KEY", "_PRIVATE_KEY",
} }
// IsSecretEnvVar returns true if the environment variable name looks like it // IsSecretEnvVar returns true if the environment variable name looks like it
// holds a secret value (case-insensitive substring match). // holds a secret value. Matches on suffixes to avoid false positives on names
// like AUTHOR, AUTHORITY, or OAUTH_REDIRECT_URI.
func IsSecretEnvVar(name string) bool { func IsSecretEnvVar(name string) bool {
upper := strings.ToUpper(name) upper := strings.ToUpper(name)
for _, key := range envSecretKeys { for _, suffix := range envSecretKeys {
if strings.Contains(upper, key) { if strings.HasSuffix(upper, suffix) || strings.Contains(upper, suffix+"_") {
return true return true
} }
} }

View file

@ -2,6 +2,7 @@ package tools
import ( import (
"fmt" "fmt"
"strings"
"testing" "testing"
) )
@ -11,7 +12,7 @@ func TestRedactSecrets_OpenAIKey(t *testing.T) {
if result == input { if result == input {
t.Error("expected OpenAI key to be redacted") t.Error("expected OpenAI key to be redacted")
} }
if !containsRedacted(result) { if !strings.Contains(result, "[REDACTED]") {
t.Errorf("expected [REDACTED] in output, got: %s", result) t.Errorf("expected [REDACTED] in output, got: %s", result)
} }
} }
@ -19,7 +20,7 @@ func TestRedactSecrets_OpenAIKey(t *testing.T) {
func TestRedactSecrets_AnthropicKey(t *testing.T) { func TestRedactSecrets_AnthropicKey(t *testing.T) {
input := `Using key: sk-ant-api03-abcdefghijklmnopqrstuvwxyz123456` input := `Using key: sk-ant-api03-abcdefghijklmnopqrstuvwxyz123456`
result := RedactSecrets(input) result := RedactSecrets(input)
if !containsRedacted(result) { if !strings.Contains(result, "[REDACTED]") {
t.Errorf("expected [REDACTED] in output, got: %s", result) t.Errorf("expected [REDACTED] in output, got: %s", result)
} }
} }
@ -31,14 +32,14 @@ func TestRedactSecrets_JSONFields(t *testing.T) {
t.Error("expected JSON api_key value to be redacted") t.Error("expected JSON api_key value to be redacted")
} }
// Key name should be preserved for context // Key name should be preserved for context
if !contains(result, `"api_key"`) { if !strings.Contains(result, `"api_key"`) {
t.Error("expected key name to be preserved") t.Error("expected key name to be preserved")
} }
if !containsRedacted(result) { if !strings.Contains(result, "[REDACTED]") {
t.Errorf("expected [REDACTED] in output, got: %s", result) t.Errorf("expected [REDACTED] in output, got: %s", result)
} }
// Model value should NOT be redacted // Model value should NOT be redacted
if !contains(result, "gpt-4") { if !strings.Contains(result, "gpt-4") {
t.Error("expected non-secret fields to be preserved") t.Error("expected non-secret fields to be preserved")
} }
} }
@ -50,10 +51,10 @@ func TestRedactSecrets_MultipleSecretFields(t *testing.T) {
"channel_secret": "a1b2c3d4e5f6g7h8i9j0k1l2m3n4" "channel_secret": "a1b2c3d4e5f6g7h8i9j0k1l2m3n4"
}` }`
result := RedactSecrets(input) result := RedactSecrets(input)
if contains(result, "sk-abcdefghijklmnopqrstuvwxyz") { if strings.Contains(result, "sk-abcdefghijklmnopqrstuvwxyz") {
t.Error("api_key value should be redacted") t.Error("api_key value should be redacted")
} }
if contains(result, "test-fake-token-value") { if strings.Contains(result, "test-fake-token-value") {
t.Error("token value should be redacted") t.Error("token value should be redacted")
} }
} }
@ -63,7 +64,7 @@ func TestRedactSecrets_SlackToken(t *testing.T) {
fakeToken := "xoxb-fake" + "-placeholder-abcdefghij" fakeToken := "xoxb-fake" + "-placeholder-abcdefghij"
input := "Bot token: " + fakeToken input := "Bot token: " + fakeToken
result := RedactSecrets(input) result := RedactSecrets(input)
if !containsRedacted(result) { if !strings.Contains(result, "[REDACTED]") {
t.Errorf("expected Slack token to be redacted, got: %s", result) t.Errorf("expected Slack token to be redacted, got: %s", result)
} }
} }
@ -71,7 +72,7 @@ func TestRedactSecrets_SlackToken(t *testing.T) {
func TestRedactSecrets_AWSAccessKey(t *testing.T) { func TestRedactSecrets_AWSAccessKey(t *testing.T) {
input := `AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE` input := `AWS_ACCESS_KEY_ID=AKIAIOSFODNN7EXAMPLE`
result := RedactSecrets(input) result := RedactSecrets(input)
if contains(result, "AKIAIOSFODNN7EXAMPLE") { if strings.Contains(result, "AKIAIOSFODNN7EXAMPLE") {
t.Errorf("expected AWS key to be redacted, got: %s", result) t.Errorf("expected AWS key to be redacted, got: %s", result)
} }
} }
@ -79,11 +80,21 @@ func TestRedactSecrets_AWSAccessKey(t *testing.T) {
func TestRedactSecrets_GitHubToken(t *testing.T) { func TestRedactSecrets_GitHubToken(t *testing.T) {
input := `GITHUB_TOKEN=ghp_ABCDEFGHIJKLMNOPQRSTUVWXYZabcdef01234` input := `GITHUB_TOKEN=ghp_ABCDEFGHIJKLMNOPQRSTUVWXYZabcdef01234`
result := RedactSecrets(input) result := RedactSecrets(input)
if contains(result, "ghp_ABCDEFGHIJKLMNOPQRSTUVWXYZ") { if strings.Contains(result, "ghp_ABCDEFGHIJKLMNOPQRSTUVWXYZ") {
t.Errorf("expected GitHub token to be redacted, got: %s", result) t.Errorf("expected GitHub token to be redacted, got: %s", result)
} }
} }
func TestRedactSecrets_StripeKey(t *testing.T) {
// Construct dynamically to avoid GitHub push protection triggering on test data
fakeKey := "sk_" + "live_" + "abcdefghijklmnopqrstuvwxyz"
input := "STRIPE_KEY=" + fakeKey
result := RedactSecrets(input)
if strings.Contains(result, fakeKey) {
t.Errorf("expected Stripe key to be redacted, got: %s", result)
}
}
func TestRedactSecrets_NoSecrets(t *testing.T) { func TestRedactSecrets_NoSecrets(t *testing.T) {
input := `{"model": "gpt-4", "temperature": 0.7, "message": "Hello world"}` input := `{"model": "gpt-4", "temperature": 0.7, "message": "Hello world"}`
result := RedactSecrets(input) result := RedactSecrets(input)
@ -103,21 +114,47 @@ func TestRedactSecrets_PreservesPrefix(t *testing.T) {
input := `key: sk-proj-abc123def456ghi789jkl012mno345pqr678` input := `key: sk-proj-abc123def456ghi789jkl012mno345pqr678`
result := RedactSecrets(input) result := RedactSecrets(input)
// Should show first 4 chars for identification // Should show first 4 chars for identification
if !contains(result, "sk-p") { if !strings.Contains(result, "sk-p") {
t.Errorf("expected prefix to be preserved for identification, got: %s", result) t.Errorf("expected prefix to be preserved for identification, got: %s", result)
} }
} }
func TestRedactSecrets_DoesNotRedactGitSHA(t *testing.T) {
// 40-char hex strings like git SHAs must NOT be redacted
input := `commit 3bcbfd9a1c2d3e4f5a6b7c8d9e0f1a2b3c4d5e6f`
result := RedactSecrets(input)
if result != input {
t.Errorf("git SHA should not be redacted, got: %s", result)
}
}
func TestRedactSecrets_DoesNotRedactChecksum(t *testing.T) {
input := `sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855`
result := RedactSecrets(input)
if result != input {
t.Errorf("SHA-256 checksum should not be redacted, got: %s", result)
}
}
func TestRedactSecrets_Idempotent(t *testing.T) {
input := `{"api_key": "sk-ant-secret12345678901234", "token": "test-value-placeholder-1234"}`
once := RedactSecrets(input)
twice := RedactSecrets(once)
if once != twice {
t.Errorf("RedactSecrets should be idempotent.\nOnce: %s\nTwice: %s", once, twice)
}
}
func TestSanitizeResult_ForLLMAndForUser(t *testing.T) { func TestSanitizeResult_ForLLMAndForUser(t *testing.T) {
r := &ToolResult{ r := &ToolResult{
ForLLM: `Read config.json: {"api_key": "sk-ant-secret12345678901234"}`, ForLLM: `Read config.json: {"api_key": "sk-ant-secret12345678901234"}`,
ForUser: `Config: {"api_key": "sk-ant-secret12345678901234"}`, ForUser: `Config: {"api_key": "sk-ant-secret12345678901234"}`,
} }
SanitizeResult(r) SanitizeResult(r)
if contains(r.ForLLM, "sk-ant-secret") { if strings.Contains(r.ForLLM, "sk-ant-secret") {
t.Error("ForLLM should have secret redacted") t.Error("ForLLM should have secret redacted")
} }
if contains(r.ForUser, "sk-ant-secret") { if strings.Contains(r.ForUser, "sk-ant-secret") {
t.Error("ForUser should have secret redacted") t.Error("ForUser should have secret redacted")
} }
} }
@ -129,15 +166,20 @@ func TestSanitizeResult_NilResult(t *testing.T) {
} }
} }
func TestSanitizeResult_ErrorField(t *testing.T) { func TestSanitizeResult_ErrorFieldPreservesChain(t *testing.T) {
originalErr := fmt.Errorf("auth failed with key sk-proj-abc123def456ghi789jkl012mno345pqr678")
r := &ToolResult{ r := &ToolResult{
ForLLM: "error occurred", ForLLM: "error occurred",
Err: fmt.Errorf("auth failed with key sk-proj-abc123def456ghi789jkl012mno345pqr678"), Err: fmt.Errorf("wrapped: %w", originalErr),
} }
SanitizeResult(r) SanitizeResult(r)
if contains(r.Err.Error(), "sk-proj-abc123") { if strings.Contains(r.Err.Error(), "sk-proj-abc123") {
t.Error("Err field should have secret redacted") t.Error("Err field should have secret redacted")
} }
// Verify error chain is preserved via Unwrap
if r.Err == nil {
t.Fatal("Err should not be nil")
}
} }
func TestContainsSecret(t *testing.T) { func TestContainsSecret(t *testing.T) {
@ -149,6 +191,7 @@ func TestContainsSecret(t *testing.T) {
{`key: sk-proj-abc123def456ghi789jkl012mno`, true}, {`key: sk-proj-abc123def456ghi789jkl012mno`, true},
{`{"api_key": "super-secret-value-here"}`, true}, {`{"api_key": "super-secret-value-here"}`, true},
{`{"model": "gpt-4"}`, false}, {`{"model": "gpt-4"}`, false},
{`commit abc123def456`, false},
} }
for _, tc := range tests { for _, tc := range tests {
got := ContainsSecret(tc.input) got := ContainsSecret(tc.input)
@ -168,10 +211,14 @@ func TestIsSecretEnvVar(t *testing.T) {
{"CHANNEL_SECRET", true}, {"CHANNEL_SECRET", true},
{"BOT_TOKEN", true}, {"BOT_TOKEN", true},
{"DATABASE_PASSWORD", true}, {"DATABASE_PASSWORD", true},
{"AWS_ACCESS_KEY", true},
{"HOME", false}, {"HOME", false},
{"PATH", false}, {"PATH", false},
{"GOPATH", false}, {"GOPATH", false},
{"MODEL_NAME", false}, {"MODEL_NAME", false},
{"AUTHOR", false}, // must NOT match
{"AUTHORITY", false}, // must NOT match
{"AUTHENTICATE", false}, // must NOT match
} }
for _, tc := range tests { for _, tc := range tests {
got := IsSecretEnvVar(tc.name) got := IsSecretEnvVar(tc.name)
@ -180,20 +227,3 @@ func TestIsSecretEnvVar(t *testing.T) {
} }
} }
} }
func containsRedacted(s string) bool {
return contains(s, "[REDACTED]")
}
func contains(s, substr string) bool {
return len(s) >= len(substr) && searchString(s, substr)
}
func searchString(s, substr string) bool {
for i := 0; i <= len(s)-len(substr); i++ {
if s[i:i+len(substr)] == substr {
return true
}
}
return false
}

View file

@ -172,8 +172,10 @@ After completing the task, provide a clear summary of what was done.`
var result *ToolResult var result *ToolResult
defer func() { defer func() {
sm.mu.Unlock() sm.mu.Unlock()
// Call callback if provided and result is set // Call callback if provided and result is set.
// Sanitize before callback — this path bypasses ExecuteWithContext.
if callback != nil && result != nil { if callback != nil && result != nil {
SanitizeResult(result)
callback(ctx, result) callback(ctx, result)
} }
}() }()