Skills neon-postgres Audit History
๐Ÿ“ฆ

Audit History

neon-postgres - 5 audits

Version comparison

Capability and finding changes across audited versions, newest first.

VersionDateResultReview itemsChange vs previous
v5 LatestAug 4, 2026, 03:32 PM 2 confirmed14External commandsNetwork access
v4 Jul 7, 2026, 12:51 AM No confirmed findings0No capability change
v3 Jul 7, 2026, 12:51 AM No confirmed findings0No capability change
v2 Jun 30, 2026, 04:18 PM No confirmed findings1Env variables
v1 Feb 24, 2026, 03:54 PM No confirmed findings0Baseline

Aug 4, 2026, 03:32 PM

Most static alerts are false positives caused by Markdown backticks, code fences, official links, and standard process.env examples. Confirmed risks are unattended remote installation, cloud-changing Neon CLI commands, and instructions to inspect .env files containing database credentials. No prompt injection or exfiltration intent was found.

1
Files scanned
385
Lines analyzed
19
Review items
0
False positives ignored

Confirmed security concerns (2)

High
Environment file access
Use MCP server or CLI to get the connection string. Store it in `.env` as `DATABASE_URL`. Read the f
The instructions call for reading or checking an .env file that holds DATABASE_URL, so database credentials may enter the agent context during setup.
High
Environment file access
If resuming setup, check what's already configured (MCP connection, `.env` with `DATABASE_URL`, depe
The instructions call for reading or checking an .env file that holds DATABASE_URL, so database credentials may enter the agent context during setup.
Capability review items (14)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High
Database connection strings
Use MCP server or CLI to get the connection string. Store it in `.env` as `DATABASE_URL`. Read the f
The setup flow instructs the agent to inspect an .env file containing DATABASE_URL, which can expose a live database credential in the agent context.
High
Database connection strings
If resuming setup, check what's already configured (MCP connection, `.env` with `DATABASE_URL`, depe
The setup flow instructs the agent to inspect an .env file containing DATABASE_URL, which can expose a live database credential in the agent context.
Medium
Ruby/shell backtick execution
Offer to inspect existing connected Neon projects or create new ones using the Neon CLI or MCP serve
The skill directs npx with automatic confirmation to install and run the latest Neon CLI, which executes unpinned remote package code and initiates account authorization.
Medium
Ruby/shell backtick execution ยท 3 occurrences
```bash
The command runs neon@latest through npx with -y, so remote unpinned code executes without an installation confirmation.
Medium
Ruby/shell backtick execution
This installs the Neon extension (for Cursor/VS Code) or MCP server (for other agents), creates an A
The described init operation installs integrations, creates an API key, and changes project configuration, giving the external command security-sensitive side effects.
Medium
Ruby/shell backtick execution
If `init` is not suitable, the individual steps can be run non-interactively:
This section explicitly introduces non-interactive installation commands, which can modify global agent configuration without a separate confirmation.
Medium
Ruby/shell backtick execution
- **MCP server:** `npx -y add-mcp https://mcp.neon.tech/mcp -g -n Neon -y -a <agent-name>`
The command runs add-mcp through npx with -y and installs a remote Neon MCP endpoint globally, creating a supply-chain and account-access risk.
Medium
Ruby/shell backtick execution
- **Agent skill:** `npx skills add neondatabase/agent-skills --skill neon-postgres --agent <agent-na
The command installs a remote agent skill through npx with automatic confirmation, so repository content is added without an explicit review step.
Medium
Ruby/shell backtick execution
Use this for local development enablement with `npx -y neon@latest init --agent <agent-name>`, VSCod
The recommended npx setup runs the latest Neon package and configures development integrations, which executes remote code and changes agent configuration.
Medium
Ruby/shell backtick execution
Reconcile the declaration from the CLI โ€” the Neon equivalent of `terraform plan` / `apply`:
The prose directs users toward Neon plan and apply commands; apply changes live cloud service and compute settings.
Medium
Ruby/shell backtick execution
neon deploy # alias for `neon config apply`
The neon deploy alias applies declared configuration to live Neon resources, so execution has material cloud-side effects.
Medium
Ruby/shell backtick execution
Because `neon checkout` applies the policy as it **creates** a branch, a fresh branch comes up with
The instructions state that neon checkout and neon deploy create or reconcile cloud resources, making command execution operationally consequential.
Audited by: codex

Jul 7, 2026, 12:51 AM

The only static finding is a documentation reference to DATABASE_URL in SKILL.md. I treated it as a false positive because the skill does not access environment variables, request secrets, or include code that could exfiltrate credentials. No prompt injection or other semantic security issues were found.

1
Files scanned
57
Lines analyzed
1
Review items
0
False positives ignored

Risk Factors

๐Ÿ”‘ Env variables (1)
Audited by: codex

Jul 7, 2026, 12:51 AM

The only static finding is a documentation reference to DATABASE_URL in SKILL.md. I treated it as a false positive because the skill does not access environment variables, request secrets, or include code that could exfiltrate credentials. No prompt injection or other semantic security issues were found.

1
Files scanned
57
Lines analyzed
1
Review items
0
False positives ignored

Risk Factors

๐Ÿ”‘ Env variables (1)
Audited by: codex

Jun 30, 2026, 04:18 PM

Static analysis flagged database connection string terminology and two weak cryptography patterns in SKILL.md. Review found documentation-only references to Neon connection URLs and PgBouncer, with no executable code, no credential collection, and no weak cryptographic implementation. No prompt injection attempt or data exfiltration behavior was found.

1
Files scanned
57
Lines analyzed
2
Review items
1
False positives ignored
Capability review items (1)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

Low
Database URL Terminology Requires Secret Handling Care
SKILL.md references DATABASE_URL and DIRECT_URL as connection string names for Prisma with Neon. This is legitimate database setup guidance, but users should avoid pasting real credentials into prompts or shared logs.
The lines describe expected environment variable names for Neon and Prisma, not code that reads or transmits secrets. The remaining concern is user handling of database credentials.
Static false positives ignored (1)

These static matches were dismissed by semantic review or matched schema-only tokens, so they are shown for transparency but do not drive the quality score.

Low
False Positive Weak Cryptography Pattern
Static analysis flagged weak cryptography at the skill description and PgBouncer guidance. The referenced lines contain product and connection pooling text, not hashing, encryption, or cryptographic API usage.
The cited lines contain plain documentation text and no code path. There is no evidence of MD5, SHA1, DES, or other weak cryptographic operations.

Risk Factors

๐Ÿ”‘ Env variables (1)
Audited by: codex

Feb 24, 2026, 03:54 PM

Static analysis flagged 4 patterns that are all false positives. The skill contains documentation-only content with no executable code. Environment variable references are configuration examples, not actual env access. Cryptographic algorithm warnings are pattern-matching errors on unrelated text.

1
Files scanned
61
Lines analyzed
0
Review items
0
False positives ignored
No confirmed security findings were recorded for this completed audit.
Audited by: claude