Snapshot: full project state

audit.py

@@ -0,0 +1,87 @@
+# Procyon — self-audit before send. Every item must PASS before an application
+# is eligible for the user's approval/send gate.
+
+import re
+
+import db
+import tailor
+
+
+def run_audit(app_id, job, email_subject, email_body, tailored_resume):
+    """Run the checklist, log results, return list of {item, result, detail}."""
+    checks = []
+
+    def add(item, result, detail):
+        checks.append({'item': item, 'result': result, 'detail': detail})
+        db.add_audit(app_id, job['id'], item, result, detail)
+
+    company = (job.get('company') or '').strip()
+    title = (job.get('title') or '').strip()
+
+    # 1. Employer name present in email (personalization signal, NOT a safety gate — the LLM
+    # often abbreviates ("PSI" for "Physical Superintelligence") or says "your team", and the
+    # subject line already carries the company, so a missing exact match is only a WARN).
+    if company:
+        if company.lower() in (email_body or '').lower():
+            add('Employer name in email', 'PASS', f'"{company}" appears in body')
+        else:
+            add('Employer name in email', 'WARN', f'"{company}" not verbatim in body (may be abbreviated)')
+    else:
+        add('Employer name in email', 'WARN', 'No company recorded — verify before send')
+
+    # 2. No leftover placeholder / other-company names
+    placeholders = ['[company]', '[name]', '{{', '}}', 'lorem', 'TODO', 'xxx']
+    found = [p for p in placeholders if p.lower() in (email_body or '').lower()]
+    add('No placeholder tokens', 'FAIL' if found else 'PASS',
+        f'placeholders: {found}' if found else 'clean')
+
+    # 3. Contact info present
+    phone = db.get_setting('resume_phone', '')
+    email = db.get_setting('resume_email', '')
+    missing = []
+    if phone and phone not in (email_body or ''):
+        missing.append('phone')
+    if email and email not in (email_body or ''):
+        missing.append('email')
+    add('Contact info in email', 'FAIL' if missing else 'PASS',
+        f'missing: {missing}' if missing else 'phone + email present')
+
+    # 4. Subject present and sane length
+    if not email_subject:
+        add('Email subject', 'FAIL', 'missing')
+    elif len(email_subject) > 120:
+        add('Email subject', 'FAIL', f'too long ({len(email_subject)} chars)')
+    else:
+        add('Email subject', 'PASS', email_subject)
+
+    # 5. Tailored resume non-empty and no fabrication markers
+    if not tailored_resume or len(tailored_resume) < 100:
+        add('Tailored resume', 'FAIL', 'too short / empty')
+    else:
+        add('Tailored resume', 'PASS', f'{len(tailored_resume)} chars')
+
+    # 6. Resume is a real deterministic resume (was "targeting header matches job" — removed:
+    # the resume is now deterministic and no longer carries a job-name header, so that check
+    # always false-warned).
+    if tailored_resume and len(tailored_resume) > 100:
+        add('Resume present', 'PASS', f'{len(tailored_resume)} chars, deterministic facts')
+    else:
+        add('Resume present', 'FAIL', 'resume missing')
+
+    # 7. ATS keyword coverage — quality signal, never blocks sending (informational only).
+    try:
+        kws = tailor.extract_job_keywords(job.get('description') or '', top_n=12)
+        cov, _kws, missing = tailor.ats_coverage(tailored_resume, kws)
+        if cov >= 0.5:
+            add('ATS keyword coverage', 'PASS', f'{int(cov*100)}% of top keywords present')
+        else:
+            add('ATS keyword coverage', 'WARN',
+                f'{int(cov*100)}% — resume may not strongly echo this role ({", ".join(missing[:5])})')
+    except Exception:
+        add('ATS keyword coverage', 'WARN', 'could not compute')
+
+    return checks
+
+
+def all_pass(checks):
+    return all(c['result'] != 'FAIL' for c in checks)