fix: P16.1 file-existence + JS DSGAI20 coverage (audit M4/L5)

dsgai_scanner_tool/build/export_semgrep.py

@@ -35,6 +35,8 @@ def render():
     for r in data["rules"]:
         if r["classification"] != "structural":
             continue
+        if r.get("match") == "file_exists":
+            continue  # file-existence rules aren't expressible as pattern-regex
         sev = SEVERITY.get(r["signal"], "INFO")
         includes = ", ".join(yq(g) for g in r["file_globs"])
         # Semgrep pattern-regex runs against whole-file text; force multiline so a

dsgai_scanner_tool/cli/dsgai_scan.py

@@ -616,7 +616,12 @@ def cmd_scan(args):
             candidates = [f for f in files
                           if glob_match(f[1], rule["file_globs"])
                           and not glob_match(f[1], rule.get("exclude_globs") or [])]
-            matches_by_rule[rule["id"]] = run_rule(rg, rule, candidates, scan_root)
+            if rule.get("match") == "file_exists":
+                # Presence of a matching FILE is the signal (e.g. an AI-ignore
+                # file exists → PASS). No content grep. Anchor at line 1.
+                matches_by_rule[rule["id"]] = {(f[1], 1) for f in candidates}
+            else:
+                matches_by_rule[rule["id"]] = run_rule(rg, rule, candidates, scan_root)
             rn = rule.get("requires_nearby") or {}
             if rn.get("pattern"):
                 # Locate the corroborating pattern (always structural — a code

dsgai_scanner_tool/dist/dsgai.semgrep.yaml

@@ -743,19 +743,6 @@ rules:
       dsgai_rule: P12.7
       confidence: medium
       framework: 'dsgai-2026-v1.0'
-  - id: dsgai-P16.1
-    languages: [generic]
-    severity: INFO
-    message: 'DSGAI16 P16.1: AI-ignore file present (PASS)'
-    patterns:
-      - pattern-regex: '(\.(copilot|ai|cursor|continue|codeium)ignore)'
-    paths:
-      include: ['.copilotignore', '.aiignore', '.cursorignore', '.continueignore', '.codeiumignore', '.vscode/settings.json', 'cursor.json', 'continue.json']
-    metadata:
-      dsgai_control: DSGAI16
-      dsgai_rule: P16.1
-      confidence: medium
-      framework: 'dsgai-2026-v1.0'
   - id: dsgai-P16.2
     languages: [generic]
     severity: INFO
@@ -997,7 +984,7 @@ rules:
     patterns:
       - pattern-regex: '(?m)(api_key[^\n]{0,15}(verify|validate)|bearer_token[^\n]{0,15}validate|Authorization[^\n]{0,15}required|@require_auth|auth_middleware|Depends\(.*auth)'
     paths:
-      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     metadata:
       dsgai_control: DSGAI20
       dsgai_rule: P20.1
@@ -1010,7 +997,7 @@ rules:
     patterns:
       - pattern-regex: '(rate_limit|RateLimiter|@throttle|slowapi|flask_limiter|express-rate-limit|@limiter\.limit)'
     paths:
-      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     metadata:
       dsgai_control: DSGAI20
       dsgai_rule: P20.2
@@ -1023,7 +1010,7 @@ rules:
     patterns:
       - pattern-regex: '(max_input_length\s*=|max_input_tokens\s*=|input[._-]truncat|validate[._-]length)'
     paths:
-      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     metadata:
       dsgai_control: DSGAI20
       dsgai_rule: P20.3
@@ -1036,7 +1023,7 @@ rules:
     patterns:
       - pattern-regex: '(prompt_injection_detect|detect_injection|llm_guard|rebuff|input_guard|nemo[._-]guard)'
     paths:
-      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     metadata:
       dsgai_control: DSGAI20
       dsgai_rule: P20.4
@@ -1049,7 +1036,7 @@ rules:
     patterns:
       - pattern-regex: '(?m)(@app\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict)|@router\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict)|app\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict))'
     paths:
-      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+      include: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     metadata:
       dsgai_control: DSGAI20
       dsgai_rule: P20.5

dsgai_scanner_tool/rules/dsgai-rules.json

@@ -1765,6 +1765,7 @@
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P16.1",
+      "match": "file_exists",
       "name": "ai-ignore-file-present",
       "notes": "Filename-existence check (not content grep). Absence in a repo with .env or secrets/ = WARN.",
       "pcre": "(\\.(copilot|ai|cursor|continue|codeium)ignore)",
@@ -2138,7 +2139,8 @@
         "*.ts",
         "*.java",
         "*.kt",
-        "*.go"
+        "*.go",
+        "*.js"
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P20.1",
@@ -2157,7 +2159,8 @@
         "*.ts",
         "*.java",
         "*.kt",
-        "*.go"
+        "*.go",
+        "*.js"
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P20.2",
@@ -2176,7 +2179,8 @@
         "*.ts",
         "*.java",
         "*.kt",
-        "*.go"
+        "*.go",
+        "*.js"
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P20.3",
@@ -2195,7 +2199,8 @@
         "*.ts",
         "*.java",
         "*.kt",
-        "*.go"
+        "*.go",
+        "*.js"
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P20.4",
@@ -2214,7 +2219,8 @@
         "*.ts",
         "*.java",
         "*.kt",
-        "*.go"
+        "*.go",
+        "*.js"
       ],
       "framework": "dsgai-2026-v1.0",
       "id": "P20.5",

dsgai_scanner_tool/rules/dsgai-rules.yaml

@@ -899,6 +899,7 @@ rules:
     control: DSGAI16
     name: ai-ignore-file-present
     classification: structural
+    match: file_exists
     signal: pass_signal
     confidence: medium
     pcre: '(\.(copilot|ai|cursor|continue|codeium)ignore)'
@@ -1119,7 +1120,7 @@ rules:
     signal: pass_signal
     confidence: medium
     pcre: '(api_key[^\n]{0,15}(verify|validate)|bearer_token[^\n]{0,15}validate|Authorization[^\n]{0,15}required|@require_auth|auth_middleware|Depends\(.*auth)'
-    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     exclude_globs: []
     framework: 'dsgai-2026-v1.0'
     description: 'API auth required (PASS)'
@@ -1130,7 +1131,7 @@ rules:
     signal: pass_signal
     confidence: medium
     pcre: '(rate_limit|RateLimiter|@throttle|slowapi|flask_limiter|express-rate-limit|@limiter\.limit)'
-    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     exclude_globs: []
     framework: 'dsgai-2026-v1.0'
     description: 'Rate limit (PASS)'
@@ -1141,7 +1142,7 @@ rules:
     signal: pass_signal
     confidence: medium
     pcre: '(max_input_length\s*=|max_input_tokens\s*=|input[._-]truncat|validate[._-]length)'
-    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     exclude_globs: []
     framework: 'dsgai-2026-v1.0'
     description: 'Input length validation (PASS)'
@@ -1152,7 +1153,7 @@ rules:
     signal: pass_signal
     confidence: medium
     pcre: '(prompt_injection_detect|detect_injection|llm_guard|rebuff|input_guard|nemo[._-]guard)'
-    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     exclude_globs: []
     framework: 'dsgai-2026-v1.0'
     description: 'Prompt injection detect (PASS)'
@@ -1163,7 +1164,7 @@ rules:
     signal: count
     confidence: medium
     pcre: '(@app\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict)|@router\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict)|app\.(post|get)\(["''][^"'']*(chat|generate|completion|infer|predict))'
-    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go']
+    file_globs: ['*.py', '*.ts', '*.java', '*.kt', '*.go', '*.js']
     exclude_globs: []
     framework: 'dsgai-2026-v1.0'
     description: 'Inference endpoint (count)'

dsgai_scanner_tool/rules/rules.schema.json

@@ -29,6 +29,7 @@
         "control": { "type": "string", "pattern": "^DSGAI[0-9]{2}$" },
         "name": { "type": "string", "pattern": "^[a-z0-9]+(-[a-z0-9]+)*$" },
         "classification": { "enum": ["structural", "value_bearing"] },
+        "match": { "enum": ["content", "file_exists"] },
         "signal": { "enum": ["fail", "warn", "pass_signal", "count", "info"] },
         "confidence": { "enum": ["high", "medium", "low"] },
         "pcre": { "type": "string", "minLength": 1 },

dsgai_scanner_tool/tests/expected-findings.yaml

@@ -5,46 +5,51 @@
 # PR-05 self-test asserts the scan reproduces this set exactly. Line numbers are
 # load-bearing — regenerate with tests/regen_expected.py after editing a fixture.
 #
-# As of PR-11 there are NO known_bug entries: the confirmed v0.2 false negative
-# (unquoted .env) and false positive (innocent webhook) are fixed and reflected
-# below. P02.9 (raw token literals) now catches keys assigned to arbitrary names.
+# Post-audit: no known_bug entries remain; DSGAI16 now PASSes via the .aiignore
+# file-existence rule (P16.1), and the JS /chat endpoint is caught (P20.5) now
+# that DSGAI20 scans *.js.
 framework: dsgai-2026-v1.0
 
 findings:
-  # DSGAI02 — credentials. P02.1 is now quote-optional (catches the unquoted .env);
-  # P02.9 catches raw token literals (sk-proj-, xoxb-, ...) by any variable name.
+  # DSGAI16 — AI-ignore file present (P16.1 file-existence → PASS) + sensitive paths listed
+  - {control: DSGAI16, rule_id: P16.1, path: .aiignore, line: 1, status: pass_signal}
+  - {control: DSGAI16, rule_id: P16.2, path: .aiignore, line: 1, status: info}
+  - {control: DSGAI16, rule_id: P16.2, path: .aiignore, line: 2, status: info}
+  # DSGAI02 — credentials (P02.1 quote-optional; P02.9 raw tokens, incl. gitignored .env)
   - {control: DSGAI02, rule_id: P02.1, path: .env, line: 3, status: fail, classification: value_bearing}
   - {control: DSGAI02, rule_id: P02.9, path: .env, line: 3, status: fail, classification: value_bearing}
   - {control: DSGAI02, rule_id: P02.1, path: config.py, line: 7, status: fail, classification: value_bearing}
   - {control: DSGAI02, rule_id: P02.9, path: config.py, line: 7, status: fail, classification: value_bearing}
   - {control: DSGAI02, rule_id: P02.7, path: good_config.py, line: 11, status: pass_signal, classification: value_bearing}
   - {control: DSGAI02, rule_id: P02.9, path: js-service/index.js, line: 11, status: fail, classification: value_bearing}
-  # Multi-language credential coverage (PR-15): P02.9 now scans *.cs/*.rs/*.rb
-  - {control: DSGAI02, rule_id: P02.9, path: polyglot/Config.cs, line: 5, status: fail, classification: value_bearing}
-  - {control: DSGAI02, rule_id: P02.9, path: polyglot/config.rb, line: 3, status: fail, classification: value_bearing}
-  - {control: DSGAI02, rule_id: P02.9, path: polyglot/config.rs, line: 3, status: fail, classification: value_bearing}
+  # DSGAI20 — unauthenticated JS /chat endpoint (now caught: *.js coverage)
+  - {control: DSGAI20, rule_id: P20.5, path: js-service/index.js, line: 13, status: fail}
   # DSGAI04 — supply chain
   - {control: DSGAI04, rule_id: P04.1, path: loader.py, line: 10, status: fail}
   - {control: DSGAI04, rule_id: P04.4, path: requirements.txt, line: 6, status: warn}
   - {control: DSGAI04, rule_id: P04.4, path: requirements.txt, line: 7, status: warn}
   # DSGAI06 — MCP / plugin
   - {control: DSGAI06, rule_id: P06.1, path: mcp_config.json, line: 4, status: fail}
   - {control: DSGAI06, rule_id: P06.5, path: server.py, line: 18, status: fail}
-  # DSGAI13 — P13.3 also matches host 0.0.0.0 in server.py under v0.2 globs
+  # DSGAI13 — P13.3 host 0.0.0.0 in server.py
   - {control: DSGAI13, rule_id: P13.3, path: server.py, line: 18, status: warn, classification: value_bearing}
+  # DSGAI02 polyglot (C#/Rust/Ruby credential coverage)
+  - {control: DSGAI02, rule_id: P02.9, path: polyglot/Config.cs, line: 5, status: fail, classification: value_bearing}
+  - {control: DSGAI02, rule_id: P02.9, path: polyglot/config.rb, line: 3, status: fail, classification: value_bearing}
+  - {control: DSGAI02, rule_id: P02.9, path: polyglot/config.rs, line: 3, status: fail, classification: value_bearing}
   # DSGAI11 — multi-tenant isolation
   - {control: DSGAI11, rule_id: P11.1, path: retriever.py, line: 12, status: fail}
   - {control: DSGAI11, rule_id: P11.1, path: retriever.py, line: 30, status: pass_signal}
   - {control: DSGAI11, rule_id: P11.2, path: retriever.py, line: 30, status: pass_signal}
   # DSGAI05 — RAG tenant-filter PASS signals
   - {control: DSGAI05, rule_id: P05.2, path: retriever.py, line: 30, status: pass_signal}
   - {control: DSGAI05, rule_id: P05.3, path: retriever.py, line: 30, status: pass_signal}
-  # DSGAI12 — database agent. P12.1 now fires only with an LLM call nearby (webhook dropped).
+  # DSGAI12 — database agent
   - {control: DSGAI12, rule_id: P12.2, path: sql_agent.py, line: 7, status: fail}
   - {control: DSGAI12, rule_id: P12.2, path: sql_agent.py, line: 8, status: fail}
   - {control: DSGAI12, rule_id: P12.2, path: sql_agent.py, line: 13, status: fail}
   - {control: DSGAI12, rule_id: P12.1, path: sql_agent.py, line: 20, status: fail}
-  # DSGAI15 / DSGAI02 — secret in system prompt (P15.1 + P02.9 on the same line)
+  # DSGAI15 / DSGAI02 — secret in system prompt
   - {control: DSGAI02, rule_id: P02.9, path: system_prompt.py, line: 8, status: fail, classification: value_bearing}
   - {control: DSGAI15, rule_id: P15.1, path: system_prompt.py, line: 8, status: fail, classification: value_bearing}
   # DSGAI14 — telemetry content capture
@@ -61,8 +66,7 @@ must_not_flag:
   - {rule_id: P02.1, path: good_config.py, reason: "secret comes from Vault, not hardcoded"}
   - {rule_id: P02.9, path: good_config.py, reason: "no raw token literal — Vault retrieval"}
   - {rule_id: P20.5, path: rate_limited_api.py, reason: "endpoint is authenticated and rate limited"}
-  - {rule_id: P12.1, path: webhook.py, reason: "innocent webhook — no LLM call nearby (fixed in PR-11)"}
+  - {rule_id: P12.1, path: webhook.py, reason: "innocent webhook — no LLM call nearby"}
 
-# Things the ruleset still misses, tracked to the PR that fixes each.
-known_false_negatives:
-  - {control: DSGAI20, rule_id: P20.5, path: js-service/index.js, line: 12, reason: "unauthenticated /chat endpoint; DSGAI20 globs exclude *.js", fixed_in: PR-15}
+# No known false negatives remain (the JS /chat endpoint is now caught by P20.5).
+known_false_negatives: []

dsgai_scanner_tool/tests/fixtures/vulnerable-app/.aiignore

@@ -0,0 +1,2 @@
+.env
+secrets/

dsgai_scanner_tool/tests/regen_expected.py

@@ -55,6 +55,9 @@ def raw_matches(rg):
             rel = os.path.relpath(f, FIXTURE).replace(os.sep, "/")
             if not glob_match(rel, r["file_globs"]):
                 continue
+            if r.get("match") == "file_exists":
+                out.add((r["id"], rel, 1))   # presence of the file is the signal
+                continue
             p = subprocess.run([rg, "--pcre2", "-n", "-o", "-e", r["pcre"], f],
                                capture_output=True, text=True)
             for line in p.stdout.splitlines():