The Agent Skills Directory

[SAFE]: The skill uses a sophisticated multi-stage pipeline with mandatory quality gates and adversarial reviews at every transition. It follows structured protocols (Cairn, Ledger, Tripwire) to ensure that subagent behavior is verified and traceable.- [SAFE]: Command execution is limited to standard development tools such as git, gh, and basic shell utilities (date, mv). These operations are used for legitimate state management and archival purposes within the designated project environment.- [SAFE]: Sensitive data access is confined to the platform's internal project directories (~/.claude/projects/). The orchestrator uses these paths for persisting run state, metadata, and audit logs according to established platform conventions.- [SAFE]: The provided Python processing script ('reconcile-noticed.py') and shell test scripts perform benign data processing, including regex-based extraction and sha256-based deduplication, using only standard library modules and built-in system tools.

build