ruflo-aidefence
Reviewed
at commit b9f3245
AI safety scanning, PII detection, prompt injection defense, and adaptive threat learning
0 upvotes
Log in to upvote or save- Made by
- ruvnet
- Pinned commit
-
b9f32453f3af62b9762406bd3787fdf621105ea1 - What we checked
-
Refuses calls to the aidefence_learn tool when the text it would teach the detector contains what looks like a key, token or password, and can only deny. A /aidefence-mod command checks a text or shows counters locally. It makes no network calls, runs no programs and only writes a small status file in your project.
Review details
Static review of the source at the pinned commit; not run. Delta review of the changes since 87d8687; the rest carries over from the review at that commit. Registers session.start (reads the project root, registers /aidefence-mod, writes .claude-flow/aidefence-mod/status.json via $.fs.write), tool.call (only when the guard option is on, the default) and command.run for aidefence-mod (status, scan <text>). The tool.call hook acts only on tools named aidefence_learn, bare or under any mcp__server__ prefix; it walks every string in the input (bounded in nodes and characters, and it refuses when the input is too large to read) and returns a deny with a fixed message when a secret pattern matches, without echoing the value. All other tool calls pass through unchanged. The status file holds counters only. No fetch, no processes, no code loading. The prompt files (one command, one agent, two skills) only call ruflo-core MCP tools. The two skills list Bash in allowed-tools but contain no commands; only the agent text suggests unpinned npx @claude-flow/cli@latest memory and hooks commands, and the agent pre-approves nothing, so those run under normal permission prompts. The aidefence_* tools themselves come from the separate ruflo-core plugin and were not reviewed. scripts/smoke.sh is a test script that nothing loads (this version only updates its expected version number). Changes in this version: version bump to 0.3.4 and a repository field in plugin.json; no hook or prompt code changed.
- What it uses
-
- Reads your prompts: no
- Sees tool calls: yes
- Blocks or changes tool calls: yes
- Reads or writes files: yes
- Runs programs on your machine: no
- Calls the model (uses tokens): no
- Sends data over the network: no
- Stores data: yes
- Shows something on screen: yes
Made this mod? Add a badge to your README
Paste this into your README. The badge says reviewed once the current version has been reviewed, and links readers to this page.
[](https://modsforclaude.com/mods/ruflo-aidefence/)