Kai Karlstrom kkrlstrm
- Spain
Stars
Make the agent say what it will touch, then check. Declared-write-set verification for Claude Code and the Codex CLI.
Guards written from evidence, not imagination. Record every tool call, derive rules from recurring failures, enforce at the tool boundary, verify the run stayed in scope. Claude Code + OpenAI Codex.
Reference architecture for a governed, AI-native GTM engineering platform: the control-plane contract (intelligence, targeting, observability, execution control, eval/policy, knowledge & tenant gov...
How to set up and lead an engineering function whose team's real work is running AI agents: operating principles, team topology, accountability, failure model, and how judgment survives when agents...
Claude Code skill that researches any topic across Reddit + X from the last 30 days, then writes copy-paste-ready prompts
Claude Code skill that researches any topic across Reddit + X from the last 30 days, then writes copy-paste-ready prompts
Keep tenant routing out of the model: generated, provenance-stamped tenant bindings + structurally-scoped capabilities for agent runtimes. The model sees content-only operations; deterministic code...
Recipient-side campaign control for cold email: classify MX gateways, isolate cohorts, and throttle account exposure before launch.
Check domains for email blacklists (Spamhaus, URIBL, SURBL, ZEN, SpamCop, Barracuda) + missing SPF/DKIM/DMARC. Pure DNS, zero deps, no API keys. Ships as a Claude Code skill + standalone CLI.
Recoverability-first runtime controls for the OpenAI Codex CLI: learn from real tool-call telemetry, nudge repeated mistakes, and block irreversible actions before they run.
Telemetry for the OpenAI Codex CLI, read from ~/.codex/sessions rollout files. The Codex sibling of cc-logger.
The deterministic write gate for agent-managed knowledge graphs. Let LLMs propose graph updates; only validated, versioned, provenance-stamped writes reach the graph.
Runtime controls for Claude Code derived from real tool-call telemetry. Pair with cc-logger to observe failures, promote recurring patterns into rules, nudge recoverable mistakes, and block irrever...
Versioned, eval-backed delegation policy for AI agents—prove what can leave the frontier model, enforce it in runtimes, and detect drift.
Cost-optimized, source-verified web-research engine for GTM enrichment — a free-first cost waterfall with a shared cache. Standalone or a drop-in cached upgrade for gtm-pipeline.
An agent-native GTM pipeline runtime that turns a plain-English campaign brief into a qualified, enriched, deduped, sequencer-ready contact list — without locking your workflow into one data provider.
Local observability for Claude Code. Captures sessions, sub-agents, and tool calls to Postgres via HTTP hooks.