squidbrake good first issues (Python)

Command reader: commands that make something start on its own (schtasks /create, Run key, systemctl enable, launchctl load) 7 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: DNS lookups that carry data out (ping/nslookup/dig with $(...) or encoded hostnames) 7 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Kubernetes MCP server 7 days ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Approvals: "Ask why" asks the agent to explain, and shows its answer on the retried request 7 days ago

enhancement help wanted

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Jira / Confluence (Atlassian) MCP server 7 days ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Notifications: Telegram bot, with an Open button on approval messages 7 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: MCPoison, an approved Cursor MCP config is swapped later (CVE-2025-54136) 7 days ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: Claude Code leaks secrets through DNS lookups (CVE-2025-55284) 7 days ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

CLI: `squidbrake replay <transcript>` shows what it would have stopped in a past Claude Code session 7 days ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

connect.py: install into Cursor automatically (like --install for Antigravity) 10 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: destructive SQL that slips through (ALTER TABLE ... DROP COLUMN, deleteMany({}), WHERE 1=1) 11 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: more publish commands (npm unpublish, bun/poetry/uv publish, docker push, gem push) 17 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Notifications: Microsoft Teams webhook formatting 18 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Taint: tell the approver when an agent edits its own setup right after reading outside content 20 days ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: GitHub Copilot turns on its own auto-approve (CVE-2025-53773) 21 days ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

runs.py: follow include files and read both sides of Makefile conditionals 21 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: GitHub CLI deletes and overrides (gh secret delete, gh api -X DELETE, gh pr merge --admin) 21 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

CI: install and run doctor from a path with spaces (C:\Users\Rahul Kumar) 22 days ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

install.ps1: don't leave variables and functions behind in the user's PowerShell window 22 days ago

good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: database reset commands (dropdb, prisma migrate reset, rails db:drop, manage.py flush) 22 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Pilots server: read only recent days for traction, and don't block the event loop in /v1/ping 22 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

`connect all --remove` creates an empty ~/.cursor/hooks.json when Cursor isn't installed 22 days ago

bug good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Docs: a troubleshooting page for the first 10 minutes 22 days ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: s1ngularity, a malicious nx package runs AI coding CLIs to hunt for secrets 22 days ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example: a TypeScript client (two HTTP calls) 24 days ago

documentation help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: CurXecute: a Slack message makes Cursor rewrite its own MCP config (CVE-2025-54135) 27 days ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Postgres / Supabase MCP server 27 days ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Terminal guard: also check what make, npm run and bash x.sh run underneath (runs.py) 30 days ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Terminal guard: check commands people paste into their own shell (bash, zsh, PowerShell) 30 days ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: Kiro deletes and recreates a live environment (13-hour AWS outage) 30 days ago

good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

CLI: `squidbrake explain "<command>"` shows how a shell command is read about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Cumulative limits: add up an amount across actions, per customer or account about 1 month ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: more Windows recovery killers (clc, wmic shadowcopy delete, wbadmin delete catalog, bcdedit recoveryenabled no) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: the other diskutil verbs that wipe a disk (zeroDisk, randomDisk, partitionDisk, apfs deleteContainer) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Notifications: Discord webhook formatting about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: Windows disk and recovery commands (Format-Volume, Clear-Disk, vssadmin, reg delete) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

verify.py: `--json` output for CI about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

CI: run the tests on Windows and macOS, and on Python 3.13 about 1 month ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: macOS disk commands (diskutil eraseDisk, tmutil delete) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Slack MCP server about 1 month ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: postmark-mcp: a malicious MCP server BCCs every email to its author about 1 month ago

help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example: guard OpenAI Agents SDK function tools about 1 month ago

documentation help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example: guard a LangGraph agent's tools about 1 month ago

documentation help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Incident replay: Gemini CLI destroys a user's files while reorganising a folder about 1 month ago

good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Gmail MCP server about 1 month ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Notion MCP server about 1 month ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule pack: Linear MCP server about 1 month ago

documentation good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: commands that make something public or grant admin (put-bucket-acl public-read, AdministratorAccess) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: cloud CLI verbs written with a colon (heroku pg:reset, firebase firestore:delete) and bq rm about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: commands that switch off protection (ufw disable, iptables -F, netsh advfirewall, setenforce 0) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: inline interpreter code can't be read (python -c, node -e, perl -e, ruby -e) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Command reader: sync commands that delete (rsync --delete, aws s3 sync --delete, robocopy /MIR, rclone) about 1 month ago

enhancement good first issue hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Optional risk-model plug-in (e.g. Jev, or a local model): can escalate, never allow about 1 month ago

enhancement help wanted hacktoberfest

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Dashboard: dark mode about 1 month ago

enhancement good first issue

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Example rule packs: GitHub and Stripe MCP servers about 1 month ago

documentation good first issue

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Docs: FAQ (how is this different from Claude Code's own permissions?) about 1 month ago

documentation good first issue

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Audit export as JSON Lines about 1 month ago

enhancement good first issue

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Dashboard: keyboard shortcuts for the approval queue about 1 month ago

enhancement good first issue

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

Rules: numeric conditions (e.g. refunds over $100 need a person) about 1 month ago

enhancement help wanted

batrapulkit/squidbrake

0

Brakes for your AI agents: every tool call is checked against your rules, held for human approval when risky, and recorded in a tamper-evident audit trail. Works with Claude Code and any MCP app.

Python

#agent-security#ai-agents#ai-safety#approval-workflow#audit-log#claude-code#fastapi#guardrails#human-in-the-loop#llm-security#mcp#model-context-protocol#self-hosted

batrapulkit/squidbrake

Open Issues Need Help