Menu
Use when auditing DOTS's Claude Code config for security issues — committed secrets, over-broad permissions, hook exfiltration, MCP risk, hidden-unicode injection. Runs tools/audit_claude_config.py.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Author a culture's armor items and swap troop equipment rosters via the generator + validation pipeline. Use when adding/revamping armor. Enforces the canonical-folder + cover-attribute rules.
Give DOTS lords lore-driven skill values and traits via the DOTS SkillSet system. Use when a canonical lord has wrong stats or a culture roster needs a balance pass.
Author or revamp a DOTS culture's armor set, troop tree, and recruitment wiring end-to-end. Use for new cultures or troop-tree revamps. Follows docs/ai-includes/new-culture-authoring.md.
Use when reviewing an external repo or article to adopt practices into DOTS — security-vet first, map novel vs duplicative, port (never install), review, commit your changes.
Structured 4-phase self-debug for failing agent runs (looping, drifting, burning tokens). Capture, diagnose, contained recovery, report. Complements /investigate (which is for code bugs); this skill is for harness/agent failures.
Incrementally fix dotnet build errors with minimal diffs, one error at a time
| name | security-scan |
| description | Use when auditing DOTS's Claude Code config for security issues — committed secrets, over-broad permissions, hook exfiltration, MCP risk, hidden-unicode injection. Runs tools/audit_claude_config.py. |
| allowed-tools | Bash, Read |
A deterministic, offline security scan of DOTS's Claude Code configuration surface (.claude/, .mcp.json, settings*.json, CLAUDE.md, AGENTS.md). Ported as a DOTS-calibrated subset of obra-style "scan your own agent config" tooling (affaan-m/ECC AgentShield), 2026-05-29. See docs/reviews/ for the review that motivated it.
/ship, or periodically alongside /skill-stocktake./adopt-external flow ends here).python tools/audit_claude_config.py # human report, all severities
python tools/audit_claude_config.py --min HIGH # only HIGH+CRITICAL
python tools/audit_claude_config.py --json # machine output (CI)
Exit code: 2 if any CRITICAL finding (CI gate), 1 on usage error, 0 otherwise.
| Category | Catches |
|---|---|
secrets | committed API keys / tokens / private keys / DB URIs (placeholder-suppressed) |
permissions | over-broad grants in settings*.json allow-lists (Bash(*), Write(*), permission-bypass flags) |
hook-exfil | hook scripts that phone home, open reverse shells, read credential stores, or tamper logs |
mcp-risk | hardcoded secrets in MCP env, auto-approve of project servers, unpinned npx -y |
prompt-injection | hidden zero-width / bidi unicode, override directives, download-then-run instructions in model-facing files |
This is not a blind port of the upstream ruleset. DOTS mandates fail-open hooks (|| true, 2>/dev/null, always exit 0) per .claude/rules/harness-facts.md, so that pattern is deliberately NOT flagged here (it is upstream). When porting any future rule, calibrate it to DOTS's conventions first.
npx -y on an MCP server is a real-but-low risk).docs/ or .claude/rules/ are dampened to INFO because security docs legitimately quote them as examples.Put audit-allow: <rule-id> in a comment on the SAME line (e.g. when a doc must quote a trigger phrase as an example). Suppressions are reported in the summary, so they stay visible.
.claude/, .mcp.json etc. relative to --root, default cwd)./deep-review or /review-codex — it audits configuration, not feature code.