Richard Clarke
banner
rsclarke.bsky.social
Richard Clarke
@rsclarke.bsky.social
Application Security
Toddlers are great pass phrase generators, my son this morning, "crocodile eating plants in the roof".
December 18, 2025 at 3:08 AM
Running `screen` remotely and getting `Cannot find terminfo entry for 'xterm-ghostty'.` on a remote host. From you local terminal;

infocmp -x xterm-ghostty | ssh <remote-host> 'tic -x -o ~/.terminfo -'
November 16, 2025 at 6:44 AM
Writing better prompts for security engineering based tasks crashoverride.com/blog/prompt...
How to Prompt LLMs for Better, Faster Security Reviews
Learn how to write effective LLM prompts that improve code security reviews, reduce false positives, and help you catch real-world vulnerabilities.
crashoverride.com
November 11, 2025 at 1:06 AM
CQLLM, generating CodeQL queries using fine tuned LLM and RAG www.preprints.org/manuscript/...
October 24, 2025 at 6:37 AM
Is this in scope for Apple's bug bounty? www.youtube.com/watch?v=XOB...
October 21, 2025 at 1:42 AM
Edge cases API developers should be aware of by dochia.dev, blog.dochia.dev/blog/http_e.... A shared responsibility model for web frameworks? Understand what is secure by default (that the framework provides) and what you must implement to help protect your applications.
Dochia - API Testing CLI | Bring Chaos with Love
Automate API testing with intelligent negative and boundary testing. Find edge cases before your users do.
dochia.dev
October 3, 2025 at 1:11 AM
Trying more to leverage AI in vulnerability discovery, findings are often reported "critical". In review, constraints are ignored that limit impact, only when challenging assumptions does it reach the same conclusion. Good for finding starting points to assess from an idea.
September 5, 2025 at 9:13 AM
How you know you're researching a service on AWS.
July 11, 2025 at 7:36 AM
From AWS VPC architecture design doc to deployed. Do you really need infrastructure as code?
June 12, 2025 at 6:29 AM
Had a fun idea to expose tasks in a Taskfile as MCP tools. If I've already got a Taskfile telling local and CI how to build, why repeat this in another markdown doc for the agent? github.com/rsclarke/mc...
GitHub - rsclarke/mcp-taskfile-server: Expose Taskfile tasks as MCP tools.
Expose Taskfile tasks as MCP tools. Contribute to rsclarke/mcp-taskfile-server development by creating an account on GitHub.
github.com
June 10, 2025 at 7:31 AM
The inverse pomodoro technique. Spend 5 minutes focused on writing your prompt, break for 25 whilst you wait for the AI.
June 4, 2025 at 1:50 AM
Accidently hitting reject all by the AI's code changes and asking it to politely recreate is like relearning why we make small changes and commit often.
June 3, 2025 at 8:39 AM
I started reading about why we might have resistance to certain tasks, which led me down a path to consider, how we might align engineering and security team motives for better dependency management, in turn fixing the vulnerabilities associated with them.
May 27, 2025 at 5:41 AM
There's a whole bunch of golang MCP server implementations, when import "ai/mcp"?
May 21, 2025 at 4:58 AM
Adopting AI? "We actually just made the tools available and people self-innovated" Phil Venables talks to Clint Gibler and provides great insight to discovering transformational use cases with AI.
May 21, 2025 at 3:37 AM
At the beginning of May the Software Security Code of Practice was launched at CyberUK 2025 with the aim to improve software security and resilience. There are 14 principles over 4 themes that technology companies are expected to implement.
May 20, 2025 at 7:58 AM