claude-oracle — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited claude-oracle (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 1 high-severity and 3 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 4 flagged
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
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.
Tool discovery. Searches 17 sources in parallel to find relevant skills, plugins, and MCP servers.
| Hook | When | Action |
|---|---|---|
| PreToolUse(EnterPlanMode) | Before planning | Searches for relevant tools before planning |
| PostToolUse(Bash) | After errors | Searches for tools that solve the error |
| Command | Description |
|---|---|
/search-oracle <query> [type] | Search for tools across all 17 sources |
search("query") — search all sourcesfind_similar_queries("query") first to check if this search was done beforesearch("query") — oracle's own search across 17 sourcessearch("error description") — find tools that address this error classget_error_solutions("error") checks if solved beforesearch("error description") — oracle finds new tools| Sibling | Value | How |
|---|---|---|
| Historian | Past searches avoid duplicate discovery | Historian checked first, oracle only runs if no history match |
| Praetorian | Cached compactions contain prior tool results | Restore before re-searching |
| Gladiator | Observations reveal tool gaps | Reflection may suggest searching for tools to fill gaps |
Smithery Registry, Glama.ai, Official MCP Registry, npm, GitHub marketplace plugins, awesome-mcp-servers, awesome-mcp-lists, awesome-claude-code (ccplugins), awesome-claude-code (jmanhype), awesome-agent-skills, Playbooks.com, SkillsMP, and more.
claude mcp add oracle -- npx claude-oracle-mcp~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.