adapter-claudecode — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited adapter-claudecode (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
The @weave_protocol/adapter-claudecode package enforces WARD.md policies inside Claude Code via the PreToolUse hook system. When Claude Code is about to use a tool, the adapter reads the active WARD.md and blocks the call if it violates the policy.
~/.claude/WARD.md they want applied to every sessionweave-claude-code init [--matcher=X] [--fail-closed] # Install hook
weave-claude-code disable # Remove hook
weave-claude-code status # Show config + active policy
weave-claude-code test <tool> [--input=JSON] # Dry-run a tool call
weave-claude-code helpWhen the hook fires:
$WEAVE_WARD_PATH — explicit override<cwd>/WARD.md — project root<cwd>/.weave/WARD.md — alternate location~/.claude/WARD.md — user-global fallbackFirst match wins. If none exists, the hook passes through silently.
Claude Code → WARD capability:
Bash → shell_exec (also scans command string heuristically)Edit, MultiEdit, Write → file_writeRead → file_readGrep, LS, Glob → file_read / file_listWebFetch → http_request (checks url)WebSearch → web_searchTask → subagentNotebookEdit → notebook_editWARD.md can use either the literal Claude tool name (e.g., Bash) or the generic capability (shell_exec). The stricter rule wins.
| Situation | Action |
|---|---|
| User wants to enforce WARD in Claude Code | weave-claude-code init |
| User wants to test their policy | weave-claude-code test <tool> |
| User wants global policy across all projects | Write ~/.claude/WARD.md |
| User wants project-specific policy | Write <project>/WARD.md |
| User wants stricter failure mode | init --fail-closed |
| User wants to uninstall | weave-claude-code disable |
| User reports Claude Code blocked unexpectedly | weave-claude-code status (shows active policy and source) |
--fail-closed): broken or missing WARD.md → block the callDefault is fail-open because broken policy files shouldn't break the user's workflow. Recommend fail-closed only for high-stakes environments (production deployment agents, etc.).
@weave_protocol/ward — the policy format being enforced@weave_protocol/hundredmen — enforces the same WARD.md on MCP servers (complementary)@weave_protocol/cli — weave init scaffolds projects with a WARD.mdweave-claude-code init so it backs up your config and stays idempotent.hundredmen_declare_intent.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.