| name | security-audit |
| description | Standardized workflow for conducting external-facing security audits of Overlord product surfaces and documenting them in root-level dated reports. |
security-audit
Use this skill when asked to perform or document a security audit for the site or a defined product surface.
Purpose
Produce a security review that is suitable for external consumption:
- Clear and concise for customer stakeholders
- Specific enough for customer security teams to assess risk
- Organized by product surface, with an executive summary first
- Explicit about the AI model used to conduct the audit
Output Location And Naming
- Save every audit in
security-audits/ at the repository root.
- Name each audit file by date using ISO format:
YYYY-MM-DD.md.
- If multiple audits are created on the same date, append a short suffix such as
YYYY-MM-DD-follow-up.md.
- Do not store security audit reports in
docs/, code-reviews/, or ad hoc folders.
Audit Workflow
1. Define scope before reviewing
State the exact scope at the top of the report:
- Product or environment reviewed
- Product surfaces included
- Audit date
- AI model used
- Known limitations, if any
If the request is ambiguous, narrow the scope to the concrete surfaces you can verify from the codebase and state that choice explicitly.
2. Review by product surface
Break findings down by product surface rather than by file list. Choose the surfaces that match the work, for example:
- Authentication and access control
- Public web pages
- Authenticated application flows
- API routes and server actions
- Database and Supabase policies
- Edge functions, background jobs, and integrations (REVIEW EACH EDGE FUNCTION TO DETERMINE IF HAS SECURITY BUILT IN, OR REQUIRES JWT AUTHENTICATION)
- Client-side secrets, tokens, and configuration
- Logging, telemetry, and error reporting
- Infrastructure and deployment configuration
Do not force empty sections. Include only surfaces that were actually reviewed.
3. Record findings precisely
For each finding, include:
- Severity:
Critical, High, Medium, Low, or Informational
- Surface
- Short title
- Why it matters
- Evidence
- Recommended remediation
Evidence should point to concrete code paths, routes, settings, or observed behavior. Avoid vague claims that cannot be traced back to the product.
4. Write for external readers
- Use direct, neutral language.
- Prefer short paragraphs over long narrative blocks.
- Explain impact in practical terms.
- Avoid internal shorthand unless it is defined.
- Call out uncertainty when evidence is incomplete.
Do not present speculation as a confirmed vulnerability.
Required Report Structure
Every audit should follow this structure:
# Security Audit - YYYY-MM-DD
∑
## Executive Summary
- Overall risk posture
- Most important findings
- Whether any immediate customer action is recommended
## Audit Metadata
- Date: YYYY-MM-DD
- AI model: <model name>
- Scope: <what was reviewed>
- Method: Code review, configuration review, manual walkthrough, or other applicable methods
- Limitations: <optional>
## Findings Overview
| Severity | Count |
| --- | ---: |
| Critical | 0 |
| High | 0 |
| Medium | 0 |
| Low | 0 |
| Informational | 0 |
## Product Surface Reviews
### <Surface Name>
#### Finding: <Short title>
- Severity: <severity>
- Why it matters: <risk summary>
- Evidence: <code path, route, behavior, or configuration>
- Recommendation: <specific >
Severity:
Why it matters:
Evidence:
Recommendation:
List meaningful controls that reduce risk, if observed.
Prioritized remediation or follow-up validation steps
Executive Summary Requirements
The executive summary must appear before detailed findings and should answer:
- What was reviewed
- Whether the current posture appears acceptable for customer evaluation
- Which risks matter most right now
- Whether the audit found issues that should block adoption or require remediation first
Keep it brief, but do not strip out the decision-useful content.
Severity Guidance
Use these labels consistently:
Critical: Likely to enable major compromise, unauthorized access, or sensitive data exposure with minimal barriers
High: Serious weakness with meaningful exploitation or business impact
Medium: Important weakness that should be addressed but is not an immediate systemic failure
Low: Limited impact issue or defense-in-depth gap
Informational: Useful observation, clarification, or positive note without direct security risk
If severity is uncertain, choose the lower defensible rating and explain the uncertainty.
Minimum Quality Bar
- Include at least one evidence point for every non-informational finding.
- Separate confirmed findings from open questions.
- Prefer a small number of substantiated findings over a long list of weak ones.
- If no meaningful issues are found, say so directly and still document reviewed surfaces and residual risk.
Example
Example output path:
security-audits/2026-05-25.md
Example metadata block:
<!-- version: 1.1.0 -->