» Tag
claude-code
18 postsClaude Code Skills Guide: Automating Your Dev Workflow
How Claude Code's Skills evolved into a directory-based standard, with frontmatter controls and a real /code-review workflow example.
Spec Forge: Adversarially Certifying Specs as Behaviorally Complete
Spec Forge is a Claude Code skill that adversarially certifies specifications as behaviorally complete before any code is written.
Claude Code Burns 33K Tokens Before It Even Reads Your Prompt
Wire-level analysis shows Claude Code sends 33K tokens before reading your prompt - 4.7x OpenCode, with 3.7x higher real-task costs.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comClaude Code's Prompt Cache Cost Depends on How You Launch It
Claude Code prompt-cache costs vary by launch path: measurements reveal a large fixed overhead and a subagent fan-out cost multiplier.
Sophos study: AI coding agents look like attackers to a SIEM
Sophos telemetry shows Claude Code, Cursor, and Codex trigger SIEM alerts for credential access and evasion during normal, benign operation.
Inside Claude Code's Multi-Agent Orchestration System
Claude Code's AgentTool orchestrates sync, async, fork, teammate and remote agents via git worktree isolation for reliable parallel AI coding.
How Cursor, Claude Code, and Codex actually load your rules
Cursor loads .mdc rules conditionally, Claude Code loads CLAUDE.md upfront, Codex merges AGENTS.md by directory depth. How each tool actually works and what rule-writing habits transfer across all three.
Rethinking Agent Autonomy: A Two-Axis Model Beyond the Single Ladder
A two-axis model — agency and orchestration — for measuring AI agent autonomy, replacing the single-axis ladder engineers have relied on.
Trigger-tree tracks which docs your AI coding agent actually reads
trigger-tree is a local, zero-token tool that reveals which docs Claude Code or Codex agents actually read — and which ones they miss.
26 Repos in 29 Days With Claude Code: The Real Failures Weren't Code
A 29-day AI-coding sprint built 26 repos and 335 pages with Claude Code — real failures were SEO cannibalization, URL drift, and buggy audit tools, not code.