Skip to content

Security: oplosy/launchtap

Security

SECURITY.md

Security Policy

This repository's security review target is immutable commit 6184bc5febd43de99ab6cc2b6f71f7f90c878bb6 unless a later audit explicitly names another revision. Plan 5 Tasks 1–7 are read-only audits; Task 8 is the only remediation task and may change source only for an independently validated finding or measured quality/performance case.

Audit and reportability policy

An alert, advisory, hypothesis, test failure, complexity metric, or timing result is not a finding by itself. A reportable item needs a reachable source-to-sink or failure path, exact path:line evidence, affected asset, attacker or failure prerequisites, impact, confidence, counterevidence, and a reproducible proof or reproduction method. Quality findings use IMPORTANT or MINOR only when they show a concrete defect, maintenance cost, drift risk, or measured resource waste; style-only preferences are excluded.

The canonical schema, severity calibration, state transitions, and regression-test requirements are in the Plan 5 finding template. Surface ownership and exclusions are in the Task 1 audit pack.

Handling sensitive information

Do not commit or paste real private keys, seed phrases, Privy verification keys, database credentials, private RPC URLs, production tokens, or other secrets into source, reports, tests, logs, prompts, or issues. Use synthetic canaries for leak-path testing and redact command output. Production deployment, governance, hosting, monitoring, and external-audit evidence remains an external prerequisite documented in backlog.md.

Scope boundary

Vendored dependencies, generated artifacts, test fixtures, local-development controls, and production-conditional configuration receive the explicit provenance, reachability, or conditional disposition defined in the Task 1 ledger. They are never silently treated as clean. A passing local or CI gate does not establish production readiness without the external controls listed in the audit pack.

There aren't any published security advisories