fix(redaction): do not redact URI passwords that run through a line break (#3619)

Author: migmartri

internal/redaction/betterleaks.go

@@ -79,7 +79,7 @@ var rulesConfig string
 // betterleaks.toml, the betterleaks version, or the handling of findings in
 // this file changes what gets redacted, so that policies can tell placeholders
 // from different rulesets apart. Version 1 is the plain default ruleset.
-const RulesetVersion = 2
+const RulesetVersion = 3
 
 var (
 	defaultScanner = sync.OnceValues(newBetterleaksScanner)

internal/redaction/betterleaks.toml

@@ -58,6 +58,19 @@ regexTarget = "match"
 # redacted.
 regexes = ['''(?i)://(?:localhost|[0-9]{1,3}(?:\.[0-9]{1,3}){3}|[a-z0-9-]+(?:\.[a-z0-9-]+)+):[0-9]{1,5}@''']
 
+[[allowlists]]
+description = "A port followed by a line break read as the password of a URI, such as a port followed by more lines up to an email address"
+targetRules = ["generic-credential-uri"]
+# The password stops at whitespace in source text, but in the JSON-encoded text
+# a line break is the escape \n, so the match runs on into later lines up to
+# the next @. Only a password that is a port, and then a line break, is kept: a
+# real password followed by a line break is still redacted. A real URI on a
+# later line has its own ://, which the password cannot run through, so it is
+# still found. Only an escape counts: an odd run of backslashes before the n, r
+# or t. A \n written in source code reaches the scanner as \\n and is not a
+# line break.
+regexes = ['''^[0-9]{1,5}(?:\\\\)*\\[nrt]''']
+
 [[allowlists]]
 description = "An elided password in documentation, such as http://user:…@example.com"
 targetRules = ["generic-credential-uri"]

internal/redaction/precision_test.go

@@ -139,6 +139,15 @@ func TestRulesConfigKeepsNonSecrets(t *testing.T) {
 			text:            "issuer: http://localhost:8000@chainloop.local\n",
 			wantDefaultRule: ruleCredentialURI,
 		},
+		{
+			// The file content of a Write tool call. The scanner sees it
+			// JSON-encoded, so the password of the match ran from the port,
+			// through the escaped line breaks, to the user part of an email
+			// address on a later line.
+			name:            "uri password running through line breaks to an email address",
+			text:            "API_URL=http://localhost:48002\nCAS_ENDPOINT=localhost:49001\nTEST_EMAIL=" + sampleEmail + "\n",
+			wantDefaultRule: ruleCredentialURI,
+		},
 	}
 
 	for _, tc := range testCases {
@@ -213,6 +222,19 @@ func TestRulesConfigStillRedactsSecrets(t *testing.T) {
 			secret:   samplePassword,
 			wantRule: ruleCredentialURI,
 		},
+		{
+			name:     "password in a connection uri on the line after a host and port",
+			text:     "API_URL=http://localhost:48002\nDSN=postgres://admin:" + samplePassword + "@db.internal:5432/app\n",
+			secret:   samplePassword,
+			wantRule: ruleCredentialURI,
+			keep:     "localhost:48002",
+		},
+		{
+			name:     "password in a uri followed by a line break and an email address",
+			text:     "API_URL=http://admin:" + samplePassword + "\nTEST_EMAIL=" + sampleEmail + "\n",
+			secret:   samplePassword,
+			wantRule: ruleCredentialURI,
+		},
 	}
 
 	for _, tc := range testCases {

pkg/attestation/crafter/materials/aicodingsession/redact_test.go

@@ -50,6 +50,15 @@ const (
 	fixtureAnthropicKey = "sk-ant-api03-sT5wsx9DwmaHZDL0dUWKNhAhULxa35sUzyLFK9" +
 		"5QBTZMDJTYn8p0J7ZQbwpYGYCQeW5eXAAGtVSmhp7UO9vxHJtSBC0xpAA"
 	fixtureRepository = "https://oauth2:" + fixtureGitHubPAT + "@github.com/example/repo.git"
+
+	// The digests of testdata/session-fp-shaped.json are not secrets, but a
+	// high-entropy hex string in a file is flagged by the repository scan.
+	fixtureDigestA = "9f8e7d6c5b4a39281706f5e4d3c2b1a0" + "f9e8d7c6f9e8d7c6b5a4938271605f4e"
+	fixtureDigestB = "9f8e7d6c5b4a39281706f5e4d3c2b1a0" + "f9e8d7c69f8e7d6c5b4a39281706f5e4"
+	// The UUID and commit ids are values the generic rules must leave alone.
+	fixtureUUID        = "3bf79921-3c03-81b6-" + "afff-cb246849866f"
+	fixtureCommitStart = "9f8e7d6c5b4a39281706f5e4d3c2b1a0" + "f9e8d7c6"
+	fixtureCommitEnd   = "e25fa85e1c4b7a9d3f082165" + "0cd7e4b39a1f8c2d"
 )
 
 // fixtureSecrets maps each placeholder in the session fixtures to the value the
@@ -61,6 +70,11 @@ var fixtureSecrets = map[string]string{
 	"__GITHUB_PAT__":                      fixtureGitHubPAT,
 	"__ANTHROPIC_API_KEY__":               fixtureAnthropicKey,
 	"__GIT_REPOSITORY_WITH_CREDENTIALS__": fixtureRepository,
+	"__SHA256_DIGEST_A__":                 fixtureDigestA,
+	"__SHA256_DIGEST_B__":                 fixtureDigestB,
+	"__SESSION_UUID__":                    fixtureUUID,
+	"__COMMIT_START__":                    fixtureCommitStart,
+	"__COMMIT_END__":                      fixtureCommitEnd,
 }
 
 // readFixture loads a session fixture with its credential placeholders resolved.

pkg/attestation/crafter/materials/aicodingsession/testdata/session-fp-shaped.json

@@ -8,7 +8,7 @@
       "version": "2.1.83"
     },
     "session": {
-      "id": "3bf79921-3c03-81b6-afff-cb246849866f",
+      "id": "__SESSION_UUID__",
       "slug": "memoized-purring-valley",
       "started_at": "2026-03-25T15:10:49.161Z",
       "ended_at": "2026-03-25T16:59:14.988Z",
@@ -18,8 +18,8 @@
       "repository": "git@github.com:chainloop-dev/chainloop.git",
       "branch": "jiparis/pfm-6984-redact-secrets-in-session",
       "work_dir": "/Users/jiparis/projects/chainloop",
-      "commit_start": "9f8e7d6c5b4a39281706f5e4d3c2b1a0f9e8d7c6",
-      "commit_end": "e25fa85e1c4b7a9d3f0821650cd7e4b39a1f8c2d",
+      "commit_start": "__COMMIT_START__",
+      "commit_end": "__COMMIT_END__",
       "commits": [
         "9f8e7d6 perf(crafter): de-duplicate AccessChk security descriptors",
         "e25fa85 Bump Helm Chart and Dagger Version"
@@ -32,7 +32,7 @@
       "lines_removed": 87,
       "files": [
         {"path": "app/controlplane/internal/service/attestation.go", "status": "modified", "attribution": "ai"},
-        {"path": "pkg/attestation/crafter/materials/aicodingsession/redact.go", "status": "created", "attribution": "ai", "session_ids": ["3bf79921-3c03-81b6-afff-cb246849866f"]}
+        {"path": "pkg/attestation/crafter/materials/aicodingsession/redact.go", "status": "created", "attribution": "ai", "session_ids": ["__SESSION_UUID__"]}
       ]
     },
     "model": {
@@ -72,7 +72,7 @@
       "main": [
         {
           "type": "user",
-          "message": {"role": "user", "content": "the sha256 digest is 9f8e7d6c5b4a39281706f5e4d3c2b1a0f9e8d7c6f9e8d7c6b5a4938271605f4e"},
+          "message": {"role": "user", "content": "the sha256 digest is __SHA256_DIGEST_A__"},
           "timestamp": "2026-03-25T15:13:37.330Z"
         },
         {
@@ -82,7 +82,7 @@
         },
         {
           "type": "user",
-          "message": {"role": "user", "content": "Pushed the attestation.\n  digest: sha256:9f8e7d6c5b4a39281706f5e4d3c2b1a0f9e8d7c69f8e7d6c5b4a39281706f5e4\n  access sha256:9f8e7d6c5b4a39281706f5e4d3c2b1a0f9e8d7c69f8e7d6c5b4a39281706f5e4\nworkflow key: 3bf79921-3c03-81b6-afff-cb246849866f\nidempotency_key=3bf79921-3c03-81b6-afff-cb246849866f\n"},
+          "message": {"role": "user", "content": "Pushed the attestation.\n  digest: sha256:__SHA256_DIGEST_B__\n  access sha256:__SHA256_DIGEST_B__\nworkflow key: __SESSION_UUID__\nidempotency_key=__SESSION_UUID__\n"},
           "timestamp": "2026-03-25T15:14:02.000Z"
         },
         {
@@ -109,6 +109,11 @@
           "type": "user",
           "message": {"role": "user", "content": "119\tconst pat = \"__GITHUB_PAT__\" // gitleaks:allow\n120\twantURI := \"https://tracker.example.com/issue/1?token=\" + pat\n"},
           "timestamp": "2026-03-25T15:14:07.000Z"
+        },
+        {
+          "type": "assistant",
+          "message": {"role": "assistant", "content": [{"type": "tool_use", "name": "Write", "input": {"file_path": "/work/e2e/.env", "content": "ENVIRONMENT=local\nBASE_URL=http://localhost:43000\nAPI_URL=http://localhost:48002\nCAS_ENDPOINT=localhost:49001\nCHAINLOOP_API_INSECURE=true\nTEST_EMAIL=john@chainloop.local\n"}}]},
+          "timestamp": "2026-03-25T15:14:08.000Z"
         }
       ]
     },