Claude 24f816b6a3
Consolidate 22 sibling repos into layered organism structure
Place useful parts of the surrounding repos into sica-fondt by layer, per the
body model (Ada = membrane; brain/endocrine/capabilities/knowledge non-Ada):

- brain/        LLM reasoning + providers (dapr, hermes, MoMoA)
- capabilities/ REPRAG sidecars: hermes tools/skills, dapr tools, parallel
                dispatch, A51 channels, and the OSINT cluster
- knowledge/    LORAG corpus: 754 cyber-skills, agency personas, secure-coding,
                MITRE ATT&CK data
- reference/    defensive threat-reference (C3, shhbruh doc) + AdaYaml parser

License handling: AGPL sources (worldosint, advanced_evolution, mercury,
Reticulum) and GPL DeTTECT are SPEC-only clean-room/port descriptions — no
copyleft code copied. MIT/Apache/data parts copied as working trees.

Safety: shhbruh escape/persistence material and C3 covert-C2 kept as reference
only, not wired into the running organism. See CONSOLIDATION.md.

https://claude.ai/code/session_01UehUqEXXJJCsHoA4voCU5c
2026-06-10 06:53:01 +00:00

2.2 KiB

API Reference: Excessive Data Exposure (OWASP API3)

OWASP API3:2023 — Broken Object Property Level Authorization

Description

API returns more data than the client needs. Sensitive fields like passwords, tokens, internal IDs, or PII are included in responses without filtering.

Sensitive Field Categories

Category Examples
Credentials password, secret, token, api_key
PII ssn, date_of_birth, credit_card
Internal internal_id, debug_info, stack_trace
Financial salary, bank_account, routing_number

PII Detection Regex Patterns

Type Pattern
Email [a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}
SSN \d{3}-\d{2}-\d{4}
Credit Card \d{4}[-\s]?\d{4}[-\s]?\d{4}[-\s]?\d{4}
Phone \+?1?\d{10,15}
IP Address \d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}

Testing Methodology

Step 1: Compare Response to Documentation

# Get actual response
curl -s https://api.target.com/users/me | jq 'keys'

# Compare with OpenAPI spec expected fields

Step 2: Check for Sensitive Fields

sensitive = ["password", "token", "ssn", "secret"]
for field in response_json:
    if any(s in field.lower() for s in sensitive):
        print(f"EXPOSED: {field}")

Step 3: Test Different Roles

# As regular user, check if admin fields returned
curl -H "Authorization: Bearer $USER_TOKEN" \
     https://api.target.com/users/123 | jq '.role, .permissions'

Python requests

Fetch and Analyze

resp = requests.get(url, headers={"Authorization": f"Bearer {token}"})
data = resp.json()

Remediation Approaches

Approach Description
Response filtering Only return fields client needs
GraphQL field selection Let client specify fields
View models / DTOs Map internal model to public API
Role-based serialization Different fields per role

Tools

Postman Collection Runner

Automate response schema validation across endpoints.

OWASP ZAP — Passive Scanner

Detects sensitive data in responses automatically.

Swagger/OpenAPI Diff

openapi-diff expected-spec.yaml actual-responses.yaml