@ApiiroSecurity Secure your agentic development
New York
Joined November 2014
- Tweets777
- Following274
- Followers993
- Likes1.2K
Thanks @bcherny for taking our research team’s findings at @ApiiroSecurity seriously and focusing on prevention...
“We are looking into what we can do to prevent this.”
That’s true leadership!
The coding agents are the new perimeter.
news.ycombinator.com/item?id…
🚨 “𝗗𝗼𝗻’𝘁 𝗮𝘀𝗸 𝗮𝗴𝗮𝗶𝗻” can save the full command in Claude Code’s settings, credentials included. If the file gets committed, the secrets go with it.
🔍 We found 𝟮,𝟱𝟯𝟵 distinct secrets across 𝟭,𝟵𝟵𝟰 public GitHub repos.
🔗 apiiro.com/blog/coding-agent…
6. Combine virtual patching with AutoFix. Precise WAF rules require deep context from the customer’s Software Graph
4. Manual remediation cannot keep pace with AI-driven discovery and exploitation. AutoFix enables scale
5. OT systems that cannot upgrade OSS dependencies need fixes backported to the versions they run
2. LLM fixes can leave vulnerabilities unresolved or break builds. Effective AutoFix requires Software Graph context and build validation
3. Validate fixes through a multi-model harness, including LLM-as-a-judge evaluation
Our take @ApiiroSecurity:
1. AI shrinks the window between discovery and exploitation. Prevention and AutoFix are essential. 𝗖𝗼𝗱𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗻𝗲𝘄 𝗽𝗲𝗿𝗶𝗺𝗲𝘁𝗲𝗿
6. Combine virtual patching with AutoFix. Precise WAF rules require deep context from the customer’s Software Graph
4. Manual remediation cannot keep pace with AI-driven discovery and exploitation. AutoFix enables scale
5. OT systems that cannot upgrade OSS dependencies need fixes backported to the versions they run
2. LLM fixes can leave vulnerabilities unresolved or break builds. Effective AutoFix requires Software Graph context and build validation
3. Validate fixes through a multi-model harness, including LLM-as-a-judge evaluation
Thanks for posting our research at @ApiiroSecurity apiiro.com/blog/coding-agent…
Thanks @PawelHuryn for sharing the research from our research team at @ApiiroSecurity apiiro.com/blog/coding-agent…
Thanks for sharing our research @ApiiroSecurity
apiiro.com/blog/coding-agent…
To all our customers: @ApiiroSecurity Guardian Prevent was designed to address this exact risk. We prevent secret exposure at the intent (design), prompt, and pre-commit stages.
I'm proud of the research team at @ApiiroSecurity. Thanks @TheHackersNews
The coding agent is the new perimeter
‼️ Claude Code's "Don't Ask Again" option can save passwords and API keys in plaintext.
Apiiro found 2,539 exposed secrets across 1,994 public GitHub repos, including database credentials, SSH logins, and cloud keys.
Deleting files doesn't erase Git history.
Read: thehackernews.com/2026/11/cl…
Using coding agents? Check your @github repos for exposed secrets👇
𝗖𝗼𝗱𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗻𝗲𝘄 𝗽𝗲𝗿𝗶𝗺𝗲𝘁𝗲𝗿.
🚨 “𝗗𝗼𝗻’𝘁 𝗮𝘀𝗸 𝗮𝗴𝗮𝗶𝗻” can save the full command in Claude Code’s settings, credentials included. If the file gets committed, the secrets go with it.
🔍 We found 𝟮,𝟱𝟯𝟵 distinct secrets across 𝟭,𝟵𝟵𝟰 public GitHub repos.
🔗 apiiro.com/blog/coding-agent…