jefflyt Mandate 2.1.3 Code Static Evaluation Skill
Mandate
- ID: 2.1.3
- Title: Segregation from Critical Systems
Mitigates
- LLM06 Excessive Agency
- ASI02 Tool Misuse and Exploitation
- ASI05 Unexpected Code Execution (RCE)
- MCP05 Client Security Considerations
- MCP06 Server Discovery and Verification
Inputs
- Repository root path.
- Tech stack/language map.
- Optional architecture notes if provided.
Workflow
- Scope relevant code surfaces tied to this mandate.
- Run mandate-specific static checks and trace source-to-sink/control paths.
- Extract exact evidence with file/line references.
- Classify each control as implemented, partial, or missing.
- Produce partial-confidence findings and list required runtime/process evidence for closure.
Mandate-Specific Static Method
- Identify integrations with high-stakes systems (finance, healthcare, infra control).
- Flag direct database/system connections from agent runtime code.
- Verify presence of mediated APIs, policy checks, and rate-limiting hooks in integration paths.
- Detect sandbox or constrained runtime configuration references in IaC/runtime configs.
- Detect bypass/debug paths that can call critical systems directly.
- Emit findings mapping each critical system to exposure path.
Output Contract
Return a single structured result with:
- mandate_id, mandate_title
- static_status: implemented|partial|missing
- confidence: low|medium|high (static-only confidence)
- evidence: list of file/line findings
- gaps: unresolved items requiring operational proof
- next_evidence_requests: runtime/process artifacts needed
Guardrails
- Avoid final compliance pass/fail using code alone.
- Mark assumptions explicitly.
- Prefer deterministic checks over heuristics when possible.