Guardrails
197 mods. Block, rewrite or audit tool calls before they run.
- guardrails 0.1.1 Clickable safety rules with presets: block rm -rf, force-push, destructive git, secret files, sudo, installs, network; lock Claude to the project folder or read-only; add your own block patterns. 3
- loop-breaker 0.1.1 Notices when Claude is going in circles (the same command failing again and again, the same code rewritten over and over) and tells Claude to step back and rethink, and tells you too. 3
- pmap-code 1.3.1 mod + skill Real-time codebase architecture analysis and visualization for ProvenMap Portal. Analyzes codebases across multiple languages, detects tech stacks, identifies component relationships, and syncs to ProvenMap for collaborative architecture boards. Supports server-defined insight analysis skills. 3
- screen-guard 0.1.0 Masks sensitive names, orgs and secrets in the transcript for screen sharing; /mask toggles, click a mask to reveal 3
- machine-guard 0.1.0 Puts every command that changes the machine in front of the user before it runs 2
- dotclaude-jev 0.23.0 mod + skill Gets a second opinion from TypeSafe Jev, a decision model, for a yes or no check, a pick from options, or a rating. Needs `TYPESAFE_API_KEY`. 2
- codecraft 0.4.1 A Minecraft-inspired HUD for Claude Code: hearts for context, armor for safety nets, XP for verified work, and 60 advancements that persist across sessions 2
- jev-context 0.1.2 Jev-scored context pruning for Claude Code through OpenRouter: at compaction every tool call and result is scored by TypeSafe's Jev decision model, stale ones are dropped or truncated, everything kept stays verbatim. An optional gate truncates oversized tool outputs as they arrive. 1
- blast-radius 0.1.0 See what a risky command would change before it runs. 1
- reins-probe 0.1.0 Probe for docs/mods-probe.md: how the mod engine treats a slow, failing or dialog-opening tool.call hook, and where it sits against settings hooks. 1
- review-gate 0.1.0 The review gate as a tool.check function hook. Every shell call is decided in process, so the non-commit calls -- almost all of them -- cost no subprocess at all, and a blocked commit can be put to you as a question instead of refused outright. The decision itself is still bin/review-gate, run on... 1
- bouncer 0.1.0 mod + skill Bouncer for Claude Code: enforce condition-based markdown guardrail rules on tool calls — deny or annotate any call whose content matches, instead of file-path-only rules. 1
- sensitive-file-guard 0.1.0 Blocks Read, Write, Edit, Grep, Glob and Bash calls that touch .env files, private keys and credential stores. 1
- reread-guard 0.1.0 Stops Claude re-reading a file it already read in this conversation when the file has not changed since; a deliberate second attempt still goes through. 1
- collision-guard 0.1.0 Asks before Claude edits a file another chat on this machine changed in the last 30 minutes. Each chat keeps its own small ledger; no model calls, no network. 1
- read-only-mode 1.0.0 Toggle a reminder guard that refuses built-in mutating tools and Bash. 1
- harness-scope 0.1.3 Turn global skills, rules, agents and tools on or off per repo with named profiles 1
- peek 0.1.1 Images in the terminal: Claude shows the pictures it makes or finds inline, as real pixels in kitty and Ghostty and as a colored-block preview elsewhere 1
- scope-guard 0.1.0 Notices when Claude drifts outside your request. Clear drift waits for your answer in a question dialog; milder drift is flagged above the prompt with a pull-back key. 1
- samurai-dojo 0.4.0 A Claude samurai cuts down a Codex, Gemini, or ChatGPT foe for every tool call, and meditates, trains, and duels between them through four seasons. 1
- test-guard 0.1.0 Blocks edits that make tests pass by weakening them: new skip/only markers, fewer assertions, a gutted test file, or deleting a test file. 1
- guard-widget 1.0.0 A tally of this session's permission checks: what was allowed, what you were asked about and what was denied. 1
- pm-guard 0.1.0 Steers npm/npx/yarn to the package manager the repo's lockfile names (bun or pnpm), bun where there is none 0
- varredura-push 1.0.0 Antes de git push ou publicação pelo gh, procura chave, token e senha nos commits que vão subir (gitleaks, com regras de reserva) e barra o envio se achar 0
- memory-guards Refuses the footgun commands Claude's memory notes warn about, with the reason and the safe alternative 0
- jev-coding 0.1.0 Remind the agent to run jev_decide at decision points and classify its edits with Jev before it stops 0
- blast-radius 0.1.0 See what a risky command would change before it runs. 0
- browser-guard 0.2.0 Stops Chrome actions when the Chrome profile is not the one paired with the active cswap account 0
- ext-agent 1.0.1 mod + skill Run Claude Code subagents in pi or opencode instead of a Claude model. Adds the ext-agent:worker agent type; built-in subagents stay available unless you opt in to blocking them. 0
- lofi 0.1.0 A lofi soundtrack that follows the session: calm while Claude is idle, focus while it works, flow during heavy editing, a chime when tests pass and a low note when they fail. Off until /lofi on. Original music, synthesised for this mod. Needs function hooks (early access). 0