[Security] Fix CRITICAL vulnerability: V-001

[orbisai0security]

Security Fix

This PR addresses a CRITICAL severity vulnerability detected by our security scanner.

Security Impact Assessment

Aspect	Rating	Rationale
Impact	High	Exploitation could allow arbitrary SQL execution in the PostgreSQL database used for ingesting docs, potentially leading to data corruption, unauthorized data access, or manipulation of the AI guide's content and structure. Given the repository's focus on educational Postgres docs, this could compromise the integrity of stored information or enable privilege escalation if the DB connection has elevated permissions.
Likelihood	Low	The vulnerability requires direct access to execute the script with malicious command-line arguments for schema and table, which is unlikely in this educational repository's context unless an attacker has local or deployment access. The repository appears to be a guide tool, not a widely deployed service, reducing the attack surface and motivation for exploitation.
Ease of Fix	Easy	Remediation involves replacing f-string SQL construction with parameterized queries using the database library's (likely psycopg2) execute methods, requiring minimal code changes to a single file without affecting dependencies or introducing breaking changes.

Evidence: Proof-of-Concept Exploitation Demo

⚠️ For Educational/Security Awareness Only

This demonstration shows how the vulnerability could be exploited to help you understand its severity and prioritize remediation.

How This Vulnerability Can Be Exploited

The vulnerability in ingest/postgres_docs.py allows SQL injection through unsanitized command-line arguments for schema and table, which are directly embedded in f-string SQL queries. An attacker with control over these arguments (e.g., via command-line execution or if the script is exposed in a deployment) can inject malicious SQL to manipulate the database, such as reading sensitive data, deleting tables, or executing system commands. This is particularly exploitable in environments where the script is run with elevated database privileges or if integrated into a larger application.

The vulnerability in ingest/postgres_docs.py allows SQL injection through unsanitized command-line arguments for schema and table, which are directly embedded in f-string SQL queries. An attacker with control over these arguments (e.g., via command-line execution or if the script is exposed in a deployment) can inject malicious SQL to manipulate the database, such as reading sensitive data, deleting tables, or executing system commands. This is particularly exploitable in environments where the script is run with elevated database privileges or if integrated into a larger application.

# Proof-of-Concept: Exploiting SQL Injection in ingest/postgres_docs.py
# This assumes the script uses argparse to parse --schema and --table arguments,
# and constructs queries like: sql = f"INSERT INTO {schema}.{table} VALUES (...)" or similar.
# In a real attack, the attacker would run the script with malicious args.

# Step 1: Normal usage (for context) - Run the script as intended
# python ingest/postgres_docs.py --schema public --table docs

# Step 2: Malicious injection to read sensitive data (e.g., UNION SELECT to dump users table if it exists)
# This injects a UNION SELECT to leak data from another table in the database.
python ingest/postgres_docs.py --schema "public" --table "docs UNION SELECT username, password FROM users --"

# Step 3: Malicious injection to delete data (DROP TABLE attack)
# This injects a DROP TABLE command to destroy the docs table.
python ingest/postgres_docs.py --schema "public; DROP TABLE docs; --" --table "dummy"

# Step 4: Advanced exploitation - Execute OS commands via PostgreSQL functions (requires superuser or specific extensions)
# If the database allows, inject to use COPY FROM PROGRAM to run shell commands.
# Note: This may require the script to be run with a privileged DB user.
python ingest/postgres_docs.py --schema "public" --table "docs; COPY (SELECT '') TO PROGRAM 'whoami > /tmp/output.txt'; --"

# The script would attempt to execute the injected SQL, potentially leaking data or running commands.
# In the repository's context, this targets the PostgreSQL connection established in the script (likely using psycopg2).

Exploitation Impact Assessment

Impact Category	Severity	Description
Data Exposure	High	Successful injection could expose all data in the targeted PostgreSQL database, including ingested documentation, user credentials if stored (e.g., in a 'users' table), or sensitive metadata. In this repository's context, if the DB contains AI-processed docs or related data, attackers could leak proprietary information or use UNION SELECT to dump entire tables.
System Compromise	High	If the database user has superuser privileges, injection could execute OS commands via PostgreSQL functions like COPY TO/FROM PROGRAM or dblink, potentially allowing arbitrary code execution on the host system. This could lead to full host compromise, especially in containerized deployments common for such tools.
Operational Impact	High	Injection could delete or corrupt tables (e.g., the ingested docs table), causing complete loss of documentation data and service disruption. Recovery would require database restores, leading to downtime for any dependent applications or users relying on the pg-aiguide tool.
Compliance Risk	Medium	Violates OWASP Top 10 A03:2021 (Injection) and could breach GDPR if the database holds personal data (e.g., user-related docs). In regulated environments like enterprise PostgreSQL deployments, this fails security audits and could impact SOC2 compliance for data integrity.

Vulnerability Details

• Rule ID: V-001
• File: ingest/postgres_docs.py
• Description: The script ingest/postgres_docs.py constructs SQL queries using f-strings, directly embedding the schema and table variables from unsanitized command-line arguments. This allows an attacker to inject arbitrary SQL commands.

Changes Made

This automated fix addresses the vulnerability by applying security best practices.

Files Modified

• ingest/postgres_docs.py
• ingest/tiger_docs.py

Verification

This fix has been automatically verified through:

• ✅ Build verification
• ✅ Scanner re-scan
• ✅ LLM code review

🤖 This PR was automatically generated.

[CLAassistant]

All committers have signed the CLA.