Verified change intelligence for developers and founders who run agents. We track consequential releases, pricing, limits, reliability, security, and compatibility—then explain what changed, what could break, and what to do next.
Registry stats loading…
We separate release-note language from the behavior, policy, price, limit, or compatibility change that matters in practice.
We look for pricing changes, migration risk, reliability regressions, security exposure, and hidden operational limits before recommending an update.
Each report ends with a practical verdict: update, wait, test first, pin a version, change configuration, or ignore the announcement.
Claude Code v2.1.273 (2026-09-15 20:23 UTC, ~16h before this pillar) is the substantive release Anthropic skipped the named notes on for v2.1.272. Headline capability: five x-claude-code-* request headers for LLM gateways, opt-in via CLAUDE_CODE_GATEWAY_HINT_HEADERS=1 (request-class, agent-type, prev-tool-durations, compaction, context-compacted). Headline operational: MCP server disconnect notification with auto-reconnect fallback pointing at /mcp. Headline security: explicit revert of a v2.1.268 deny-rule fix on eval/env-C unanalyzable Bash patterns, with the new behavior being prompt rather than deny (legitimate time -p make build use cases were being denied). Headline reliability: long-session responsiveness fix (hook progress and sub-agent activity no longer re-process the whole conversation on every update). Plus ten additional security/policy fixes, one context-meter double-count fix, an /login+/upgrade+/extra-usage prompt-cache preservation fix, three Bedrock/Vertex/Foundry auth-reporting fixes, default auto mode on Bedrock/Vertex/Foundry switched to local classifier with CLAUDE_CODE_AUTO_MODE_SERVER=1 opt-in, three Artifact fixes, and roughly 28 additional operational fixes spanning Claude Tag, Claude Code on the web, VS Code, settings, and /bug/feedback reports. 52 release-note items total.
We do not rewrite launch posts. We connect primary sources, reproducible tests, scanner evidence, Registry records, and technical guides to the decision in front of you.
Pricing, policy, security, compatibility, and incident claims start with official documentation, repositories, advisories, or status pages. Community reports are labeled as experience, not proof.
Field reports name the version, environment, method, inputs, and limitations. If we only compared documentation, we say so.
The skills and MCP Registry remains free. Its source links, scan states, provenance, and explicit evidence gaps help test whether an ecosystem claim survives contact with real artifacts.
Durable technical guides and inspectable workflow manifests turn a news event into implementation work you can adapt and validate in your own environment.
A guide to dependency review, supply-chain risk triage, and remediation workflows.
A practical walkthrough of Model Context Protocol server structure, authentication, and tool registration.
A technical guide to turning SEO findings into concrete implementation work.
Codex, Claude Code, Gemini CLI, Cursor, OpenClaw, Hermes, MCP, models, memory, orchestration, and agent infrastructure—prioritized by cost, limits, reliability, security, compatibility, and practical capability.
Weekday impact reports, a weekly Before You Update brief, and tested comparisons when the evidence supports them. No quota filler.
The intelligence index is the canonical feed. Existing articles and payload URLs remain available while new coverage follows the evidence-first format.
Open the intelligence index