AGGIB

AGGIB/Stroq

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

24 good first / help-wanted issues · TypeScript · last activity Oct 13, 2026

16 stars 4 forks 16 watchers TypeScript Apache License 2.0

agent-security ai-agents ai-security claude-code codex coding-agents cursor developer-tools firewall hooks llm-security mcp prompt-injection secret-scanning typescript

View on GitHub Website

24 Open Issues Need Help Last updated: Oct 13, 2026

Open Issues Need Help

View All on GitHub

Windsurf: record the real Cascade hook contract for one measured release 15 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

mcp --strict: end-anchored rules, batches, forwarded-message limits, add --strict, exposure, process-group kill 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

attack: say which steps of an attack a given host can actually see (--host, docs/HOSTS.md) 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: an update of a skill or MCP server cannot be undone 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: recover from a crash between the client-config write and the lock write 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: add succeeds with a symlinked ~/.stroq and the pin then refuses forever 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: write the same stable Stroq entry path as init, not process.argv[1] 16 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

OpenClaw: record the real before_tool_call and after_tool_call contract for one measured release 18 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Copilot CLI: record the real preToolUse and postToolUse hook contract for one measured release 18 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

sent --try: scan a built-in sample session with a made-up key 19 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

doctor: add --json with a stable, versioned report 19 days ago

enhancement help wanted good first issue

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

release: after the approval, install the published package in an empty folder and run it 20 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Cursor: record and replay the Write/Delete hook contract for one measured release 21 days ago

enhancement help wanted triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

tests: make the Python oracle reproducible on macOS and report its interpreter 21 days ago

bug help wanted triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

bench --latency: show retained and dropped sample counts in terminal output 21 days ago

enhancement help wanted good first issue triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

prove: correct the advice about detecting host upgrades 21 days ago

bug help wanted good first issue triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

docs: update the website command reference after the new CLI features 21 days ago

documentation help wanted good first issue triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

doctor: report installed skill integrity without modifying the installation 21 days ago

enhancement help wanted triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: sandbox the first capability probe of a vetted MCP server 21 days ago

enhancement help wanted triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

Safe Install: add, update and remove locally confirmed Node MCP servers 21 days ago

enhancement help wanted triaged

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

bench --latency: measure what a hook call costs (p50/p95/p99) 22 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

stroq prove for the Codex CLI 22 days ago

enhancement help wanted

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

harden undo leaves the empty folder apply made, and a failed apply leaves an empty manifest 22 days ago

bug good first issue

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript

harden: recognise single-dash password flags (mysql -p, redis-cli -a, sshpass -p, docker login -p) 22 days ago

enhancement good first issue

16

Find known secrets in your AI coding agent sessions (Claude Code, Codex, Cursor), then guard the next one with local hooks that trace risky commands to the content that dictated them. Open source, no telemetry.

TypeScript

#agent-security#ai-agents#ai-security#claude-code#codex#coding-agents#cursor#developer-tools#firewall#hooks#llm-security#mcp#prompt-injection#secret-scanning#typescript