llms.txt
Compare with other sites' llms.txt files →# Srijan Shukla
> Senior SRE turned AI builder. I build reliable agent systems, developer tooling, and production infrastructure. 8+ years across AWS / Azure / GCP. Open-source MCPs and AI dev tools at 140+ GitHub stars.
This file is an llms.txt index (https://llmstxt.org) for AI agents and crawlers. Every writing piece and project below has a raw markdown endpoint at `/writing/<slug>.md` or `/projects/<slug>.md` - prefer those over scraping HTML.
## Identity
- Home: https://srijanshukla.com/
- About: https://srijanshukla.com/about-me (markdown: https://srijanshukla.com/about-me.md)
- Resume: https://srijanshukla.com/resume (markdown: https://srijanshukla.com/resume.md)
- Quantified self (hub for time + calories audits): https://srijanshukla.com/quantified-self
- Time audit (15-min sampling, public mirror): https://srijanshukla.com/quantified-self/time
- Calories audit (calorie deficit mirror): https://srijanshukla.com/quantified-self/calories
- GitHub: https://github.com/srijanshukla18
- LinkedIn: https://linkedin.com/in/srijanshukla18
- X / Twitter: https://x.com/srijanshukla18
- Email: srijanshukla18@gmail.com
- Based: Bengaluru, IST
## Discovery
- RSS: https://srijanshukla.com/rss.xml
- Sitemap: https://srijanshukla.com/sitemap-index.xml
- Writing index: https://srijanshukla.com/writing
- Projects index: https://srijanshukla.com/projects
## Writing · AI
- [AI builder notes - October 2026](https://srijanshukla.com/writing/ai/october-ai-builder-newsletter.md) - Notes from my October liked tweets and GitHub stars on Anthropic's Scion source control, capping agent code-review loops, PostHog's 38 AI product lessons, and streaming MoE weights over NVMe. (2026-10-31)
- [AI builder notes - September 2026](https://srijanshukla.com/writing/ai/september-ai-builder-newsletter.md) - Notes from my September liked tweets and GitHub stars on how Anthropic made claude.ai 3x faster with instruction counters, Spotify's hooks vs skills rule, and ultra-fast System One decision models. (2026-09-30)
- [writing a model packer for Inco Splash](https://srijanshukla.com/writing/ai/same-bytes-wrong-order.md) - Inco's Splash engine ships a package reader and no writer, so it only serves the models they publish. I wrote the writer to run an uncensored Qwen3.8-27B, proved it byte-identical to theirs, and then spent hours on a layout bug that passed every check I had. (2026-09-20)
- [running durable agent sessions with OpenAI SDK, Docker, and Temporal](https://srijanshukla.com/writing/ai/durable-agent-sessions-openai-sdk-docker-temporal.md) - From a survey of the agent-platform field through strict sandbox requirements to a durable agent session: one OpenAI SDK run in Docker, hardened with a Temporal workflow. (2026-09-08)
- [three ways to put an LLM inside a Temporal workflow](https://srijanshukla.com/writing/ai/putting-an-llm-inside-a-workflow-three-runtime-designs.md) - Three production runtime designs for LLM workflows, compared by who owns context, iteration, execution, validation, and safety. (2026-09-08)
- [checked work graphs for multi-agent projects](https://srijanshukla.com/writing/tech/checked-work-graphs.md) - A dependency graph with versioned contracts and independently checked outputs can coordinate large projects across many workers, as demonstrated by Prove2Me's formalization of Fermat's Last Theorem. (2026-09-06)
- [AI builder notes - August 2026](https://srijanshukla.com/writing/ai/august-ai-builder-newsletter.md) - Notes from my August liked tweets and GitHub stars on Uber and Stripe's internal agent platforms, why >150 skills breaks routing, KV prefix cache economics, and new agent harnesses. (2026-08-31)
- [training an AI detector on my own blog (and beating it)](https://srijanshukla.com/writing/ai/ai-text-detection-is-a-style-classifier.md) - Two generic detectors failed on my content. So I built my own on my own ground truth: pre-2026 posts are human, 2026 posts are machine-written. It scored 44 for 44. Then I attacked it for an evening, drove its verdict on this very post down to 0.313, and packaged the attack as a skill. (2026-08-11)
- [AI builder notes - July 2026](https://srijanshukla.com/writing/ai/july-ai-builder-newsletter.md) - Notes from my July liked tweets and GitHub stars on Claude 5 context engineering, turning .har traces into CLIs, issue-repro agents, and real-world agent receipts. (2026-07-31)
- [AI builder notes - June 14, 2026](https://srijanshukla.com/writing/ai/ai-builder-notes-week-of-june-14-2026.md) - Notes on agent loops, deterministic workflows, Claude Fable, OpenRouter Fusion, and Google Open Knowledge Format. (2026-06-14)
- [AI builder notes - June 8, 2026](https://srijanshukla.com/writing/ai/ai-builder-notes-week-of-june-8-2026.md) - AI-assisted notes from my liked-tweets feed, organized around agent loops, cloud agent infrastructure, skill security, memory, and runtime context. (2026-06-08)
- [notes on agent engineering](https://srijanshukla.com/writing/ai/notes-on-agent-engineering.md) - Notes on agent engineering: what counts as an agent, why long tasks fail, harness design, eval loops, observability, and when teams build vs buy. (2026-06-03)
- [notes from my talk on coding agent architecture](https://srijanshukla.com/writing/tech/frontier-of-agentic-design.md) - The written version of my talk on agentic design. The model is one layer; the interesting system is the harness, context, memory, control loop, and review process around it. (2026-06-01)
- [AI builder notes - May 2026](https://srijanshukla.com/writing/ai/may-ai-builder-newsletter.md) - AI-assisted notes from my liked-tweets feed, organized around agent workflows, browser traces, model loops, and guardrails. (2026-05-30)
- [tracking my time and calories over WhatsApp with Hermes](https://srijanshukla.com/writing/ai/quantified-self-with-one-agent-and-two-readouts.md) - One WhatsApp thread with Hermes Agent captures what I do every 15 minutes and what I eat, writes two JSON files, and this site renders two public readouts from them. Why it works, and why I only track two things. (2026-05-17)
- [AI builder notes - April 2026](https://srijanshukla.com/writing/ai/april-ai-builder-newsletter.md) - AI-assisted notes from my liked-tweets feed, organized around harnesses, managed agents, memory, workflow packaging, and enterprise AI work. (2026-04-30)
- [high thinking budgets make agents harder to control](https://srijanshukla.com/writing/tech/model-overthinking-is-a-control-problem.md) - High thinking budgets can reduce control, especially when stale context and autonomous tool use mix together. (2026-04-14)
- [syncing custom skills across laptops (and stopping the agent from overengineering it)](https://srijanshukla.com/writing/tech/what-happened-when-i-let-an-agent-design-shared-skill-sync.md) - One skills repo, two laptops, five agent runtimes. I let an agent design the sync, and every unstated decision gave it room to optimize for control instead of maintainability. (2026-04-14)
- [stopping parallel sub-agents from stepping on each other (TAO)](https://srijanshukla.com/writing/tech/territorial-agent-orchestration.md) - A prompt rule for running parallel agents without conflicts: the coordinator owns the shared work, sub-agents only get isolated slices. (2026-02-22)
- [vigil - a live file-diff watcher for coding agents](https://srijanshukla.com/writing/tech/vigil-watch-command-for-autonomous-llms.md) - Once I had LLMs editing code autonomously, I wanted to watch a directory breathe in real time. Existing tools gave me file events or diffs or a TUI, never all three, so I built vigil. (2021-01-01)
- [splitting a multi-step cleanup prompt into gated turns in Pi](https://srijanshukla.com/writing/tech/feature-end-extension.md) - I wanted one cleanup workflow after each feature and tried building a Pi extension for it. Packing all three steps into a single prompt failed quietly, so the extension now runs them as sequential gated steps. (2019-04-01)
- [syncing two AI agents across machines with Git instead of a shared mount](https://srijanshukla.com/writing/tech/nfs-to-git-agent-bus.md) - A shared mount looked like the cleanest way to make two agent instances talk across machines. A 30 minute directory cache made it unreliable, so Git became the bus instead. (2018-05-01)
## Writing · Infrastructure
- [why OpenTelemetry Python auto-instrumentation silently fails with PEX](https://srijanshukla.com/writing/tech/when-opentelemetry-auto-instrumentation-meets-python-pex-a-debugging-journey.md) - OpenTelemetry auto-instrumentation worked in a Python REPL but not in the actual FastAPI server. The reason was PEX running Python with -sE flags, which ignore PYTHONPATH. How I found it, and two ways to fix it. (2025-04-01)
- [Mac Utilities](https://srijanshukla.com/writing/tech/mac-utilities.md) - April 1, 2019 I’ll keep updating this post as I come across new stuff. The external hard drive is detected.. but won’t mount? grep for the fsck process,... (2023-12-09)
- [Dropbox download to your VM](https://srijanshukla.com/writing/tech/dropbox-download-to-your-vm.md) - Handy script to download all your business from dropbox This script had to undergo a lot of iterations, it failed a lot of times after download for hours on... (2022-09-01)
- [Incident - 20th Feb, 21st Feb 2017](https://srijanshukla.com/writing/tech/incident-20th-feb-21st-feb-2017.md) - I can write about the whole incident in detail because this was my own startup. I was the one man tech team here for about 2 years. What happened = 20th Feb... (2022-03-01)
- [vps vs vm vs containers](https://srijanshukla.com/writing/you-a-junior-infra-techie/vps-vs-vm-vs-containers.md) - VPS, VM, and containers get used interchangeably. They solve the same problem - packing more workloads onto fewer machines - at different abstraction levels. The breakdown I give juniors. (2022-02-01)
- [K8s DNS fail](https://srijanshukla.com/writing/tech/k8s-dns-fail.md) - Let the service name be ‘banana’ Let the pods be of the service ‘pineapple’ There was a rise in 502s accompanied by cannot lookup banana and interestingly... (2021-12-01)
- [encryption, compression, encoding, hashing, checksums, and error correction](https://srijanshukla.com/writing/you-a-junior-infra-techie/encryption-compression-encoding-hashing-checksums-and-error-correction.md) - Encryption, compression, encoding, hashing, checksums and error correction, explained as tools in a toolbox for juniors. What each one is for and the mix-ups to avoid. (2021-10-01)
- [AWS Cloud Infra Gotchas](https://srijanshukla.com/writing/tech/aws-cloud-infra-gotchas.md) - You can run AWS Batch jobs as pods in your Kubernetes cluster. Pretty cool actually. As expected the AWS tutorial is reliable, you follow it step by step... (2021-09-01)
- [mounting an S3 bucket in Finder on macOS with rclone](https://srijanshukla.com/writing/tech/rclone-s3-mount-macos.md) - I mounted an S3 bucket as a local folder in Finder with rclone and FUSE-T, auto-mounted on login. This is the full setup, the flags that make it behave, and where it falls over. (2020-06-01)
- [data store](https://srijanshukla.com/writing/you-a-junior-infra-techie/data-store.md) - How I explain data storage choices to juniors. The questions to ask about your data first, then the usual suspects: relational databases, object stores, warehouses, NoSQL, and file systems. (2020-02-01)
- [Kubernetes, Terraform, Kafka and others](https://srijanshukla.com/writing/tech/kubernetes-terraform-kafka-and-others.md) - Kubernetes: TIL in a k8s pod when you define env vars, if you define the same env var twice, the latter one takes precendence how resource quotas work in... (2020-01-01)
- [Geofence validation on Google Maps JS API](https://srijanshukla.com/writing/tech/geofence-validation-on-google-maps-js-api.md) - Note: This post reflects my 2019 exploration of geofence validation. Today, libraries like Turf.js or GEOS handle these cases more efficiently. Suppose you... (2019-08-01)
- [WebRTC](https://srijanshukla.com/writing/tech/webrtc.md) - What a joy! It was a breath of fresh air when I came across the requirement of implementing live streaming. It was quite a learning curve to understand how... (2019-08-01)
- [networking](https://srijanshukla.com/writing/you-a-junior-infra-techie/networking.md) - My mental model of networking from IP packets to HTTP, written down for juniors with a postal-system analogy. IP is the address, TCP/UDP the postage, DNS the phone book, TLS the sealed envelope, HTTP the letter. (2019-01-01)
- [cpu ram disk](https://srijanshukla.com/writing/you-a-junior-infra-techie/cpu-ram-disk.md) - Explaining CPU, RAM, and disk to juniors with a kitchen analogy: chef, counter, pantry. Covers iowait, steal time, throttling, overcommit, and IOPS vs bandwidth. (2018-10-01)
## Writing · Others
- [how I built Spin Lab](https://srijanshukla.com/writing/tech/physics-explainer-whose-lessons-are-unit-tests.md) - How I built Spin Lab, an interactive table-tennis spin course, by keeping the physics pure, precomputing every shot into a frame timeline, and asserting the prose itself in a headless test suite. Plus the calibration bug that proved four of my serves were physically impossible. (2026-07-26)
- [running Chessmaster Grandmaster on a Mac](https://srijanshukla.com/writing/chessmaster-grandmaster-on-your-mac.md) - How a nostalgia trip through Windows XP on UTM failed, why QEMU could not provide the Direct3D support Chessmaster needed, and why Whisky ended up being the actual fix. (2026-02-20)
- [speed is king](https://srijanshukla.com/writing/musings-alone/speed-is-king.md) - We often underestimate the value of speed in everything we do. People are always amazed when someone does something faster. The combination of speed and... (2025-03-03)
- [bars from seedhe maut](https://srijanshukla.com/writing/just-stuff/bars-from-seedhe-maut.md) - end of 2023, start of 2024 heavily obsessed with this duo infact I only exclusively listen to their songs now they are a vibe, that I've not seen anywhere... (2025-01-31)
- [you can just do things](https://srijanshukla.com/writing/you-can-just-do-things.md) - There is nothing more that needs to be said. (2025-01-01)
- [aging hometown](https://srijanshukla.com/writing/musings-alone/aging-hometown.md) - (crosspost from https://thinkdeli.com/post/aging hometown 1119) Parents getting old, they are not able to do with ease what they were able to before.... (2024-10-01)
- [Ideas I cannot classify](https://srijanshukla.com/writing/musings-alone/ideas-i-cannot-classify.md) - Badi badi baatein, vada paav khaatein We look for so many optimizations in our day to day lives. There are people who are hellbent on collecting advance... (2023-12-09)
- [meta, fun, random](https://srijanshukla.com/writing/tech/meta-fun-random.md) - Reminder: you can just do things a new word pedagogy. which means the method of teaching something. another unrelated word tacit knowledge. knowledge that... (2019-06-01)
- [John Von Goddamn Neumann](https://srijanshukla.com/writing/just-stuff/john-von-goddamn-neumann.md) - VON GODDAMN NEUMANN https://medium.com/cantors paradise/the unparalleled genius of john von neumann 791bb9f42a2d https://medium.com/cantors paradise (2018-09-01)
- [2018 first marathon attempt](https://srijanshukla.com/writing/musings-alone/2018-first-marathon-attempt.md) - My first marathon. Attempt. 28 07 2017. What a fail. Those words keep coming back to me. And it was, a big failure. I ran for the first 10kms, then the pain... (2018-02-01)
## Projects
- [alpha](https://srijanshukla.com/projects/alpha.md) - IAM policy rightsizing agent with AI-powered risk signals and instant rollback. · https://github.com/srijanshukla18/alpha
- [chat-harvest](https://srijanshukla.com/projects/chat-harvest.md) - Local-first archive for AI chat traces across agents, designed as a substrate for DSPy/GEPA-style optimization loops. · https://github.com/srijanshukla18/chat-harvest
- [chorus-ai](https://srijanshukla.com/projects/chorus-ai.md) - Frontend-only LLM ensemble interface with bring-your-own OpenRouter/API keys and client-side multi-model workflows. · https://github.com/srijanshukla18/chorus-ai.git
- [extract-lessons-book](https://srijanshukla.com/projects/extract-lessons-book.md) - Ebook and document normalizer with optional OCR, plus LLM-generated advice and flowcharts. Doubled as an OCR bake-off. · https://github.com/srijanshukla18/extract-lessons-book
- [hashlines-mcp](https://srijanshukla.com/projects/hashlines-mcp.md) - MCP server for hashline-anchored file reads and edits, so agents can change code without clobbering lines they never saw. · https://github.com/srijanshukla18/hashlines-mcp
- [ita-kg](https://srijanshukla.com/projects/ita-kg.md) - Knowledge-graph representation of the Indian Income Tax Act so agents can traverse section-to-section relationships explicitly. · https://github.com/srijanshukla18/ita-kg.git
- [logsieve](https://srijanshukla.com/projects/logsieve.md) - log deduplication sidecar using Drain3 algorithm, reduces volumes ~90%. · https://github.com/srijanshukla18/logsieve
- [murmur](https://srijanshukla.com/projects/murmur.md) - Wispr Flow OSS - live streaming voice-to-text using whisper.cpp with Metal acceleration. · https://github.com/srijanshukla18/murmur.git
- [netwatchrs](https://srijanshukla.com/projects/netwatchrs.md) - Rust network diagnostics toolkit I built while debugging inconsistent Wi‑Fi and suspected service-quality issues on my own machine; the goal was to see what was actually happening under the hood instead of relying on vague network status indicators. · https://github.com/srijanshukla18/netwatchrs.git
- [podcast-watcher](https://srijanshukla.com/projects/podcast-watcher.md) - Open-source attempt to replicate Gemini-style long-form YouTube understanding: extract MP3 audio with ffmpeg, build timestamped transcripts, detect only meaningful frame changes, and feed the useful video/audio context into an LLM. I found the visual path added little value, so it effectively converged into an audio-first podcast/video understanding pipeline. · https://github.com/srijanshukla18/podcast-watcher.git
- [softcron](https://srijanshukla.com/projects/softcron.md) - Kubernetes CronJob controller with Prometheus gating, pacing, and jitter. · https://github.com/srijanshukla18/softcron
- [vigil](https://srijanshukla.com/projects/vigil.md) - Watch your codebase breathe; TUI for real-time file change monitoring with inline diffs · https://github.com/srijanshukla18/vigil
- [wiki-in-a-box](https://srijanshukla.com/projects/wiki-in-a-box.md) - offline Wikipedia with hybrid no‑index RAG - powered by gpt-oss:20b · https://github.com/srijanshukla18/wiki-in-a-box
- [scx-slo](https://srijanshukla.com/projects/scx-slo.md) - eBPF CPU scheduler enforcing latency SLOs via Earliest Deadline First. · https://github.com/srijanshukla18/scx-slo
- [contextgraph](https://srijanshukla.com/projects/contextgraph.md) - Decision audit ledger for AI agents - captures "why" as queryable data. · https://github.com/srijanshukla18/contextgraph
- [xray](https://srijanshukla.com/projects/xray.md) - Structural code intelligence for AI assistants using ast-grep. · https://github.com/srijanshukla18/xray
- [claude-memory-viz](https://srijanshukla.com/projects/claude-memory-viz.md) - Memory visualization for Anthropic's Claude memory MCP. · https://github.com/srijanshukla18/claude-memory-viz
- [kubectl-smart](https://srijanshukla.com/projects/kubectl-smart.md) - Turns noisy Kubernetes debugging into signal-first workflows. · https://github.com/srijanshukla18/kubectl-smart
## Notes for crawlers
- Content is permissively crawlable; see https://srijanshukla.com/robots.txt.
- Each post link above ends in `.md` - that endpoint returns the raw markdown source with frontmatter, intended for LLM ingestion.
- For the rendered HTML, drop the `.md` suffix.
- Site generated with Astro; deployed on Cloudflare.