<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0">
  <channel>
    <title>AI Programming Manual</title>
    <link>https://aiprogrammingmanual.com/</link>
    <description>Latest articles, guides, and news from AI Programming Manual.</description>
    <language>en-US</language>
    <lastBuildDate>Tue, 18 Aug 2026 00:00:00 +0000</lastBuildDate>
    <item>
      <title>OpenAI&apos;s Assistants API Shuts Down August 26: The Migration Checklist</title>
      <link>https://aiprogrammingmanual.com/news/assistants-api-shutdown-migration.html</link>
      <guid>https://aiprogrammingmanual.com/news/assistants-api-shutdown-migration.html</guid>
      <pubDate>Tue, 18 Aug 2026 00:00:00 +0000</pubDate>
      <description>OpenAI removes the Assistants API on August 26, 2026. Map assistants, threads, and runs onto the Responses and Conversations APIs, and skip the prompt objects that shut down 96 days later.</description>
    </item>
    <item>
      <title>Claude Code&apos;s Subagent Forking Default: What Changes for Your Multi-Agent Setup</title>
      <link>https://aiprogrammingmanual.com/news/subagent-forking-default.html</link>
      <guid>https://aiprogrammingmanual.com/news/subagent-forking-default.html</guid>
      <pubDate>Mon, 17 Aug 2026 00:00:00 +0000</pubDate>
      <description>Claude Code 2.1.232 makes subagent_type: &quot;fork&quot; inherit the full parent conversation and prompt cache by default. What that means for cost, context leakage between subagents, and existing custom subagent_type configs.</description>
    </item>
    <item>
      <title>Prompt Cache Hit Rate: Why Your Cached Prompt Isn&apos;t Caching</title>
      <link>https://aiprogrammingmanual.com/articles/prompt-cache-hit-rate.html</link>
      <guid>https://aiprogrammingmanual.com/articles/prompt-cache-hit-rate.html</guid>
      <pubDate>Sun, 16 Aug 2026 00:00:00 +0000</pubDate>
      <description>Provider-side prompt caching is a prefix match that fails silently. Find the invalidator in your prompt assembly, place breakpoints at stability boundaries, and confirm the fix from the response usage fields.</description>
    </item>
    <item>
      <title>Your Local Claude Code Sessions Are Now Retrievable by Your Organization</title>
      <link>https://aiprogrammingmanual.com/news/claude-code-local-sessions-compliance-api.html</link>
      <guid>https://aiprogrammingmanual.com/news/claude-code-local-sessions-compliance-api.html</guid>
      <pubDate>Sat, 15 Aug 2026 00:00:00 +0000</pubDate>
      <description>Anthropic&apos;s Compliance API now returns transcripts of Cowork and Claude Code sessions that run on users&apos; own machines, in beta for Claude Enterprise organizations. What lands in the transcript, who can retrieve it, and which local habits to change.</description>
    </item>
    <item>
      <title>Imagen 4 Shuts Down August 17: Migrate the Model ID Without Breaking Image Output</title>
      <link>https://aiprogrammingmanual.com/news/imagen-4-shutdown-migration.html</link>
      <guid>https://aiprogrammingmanual.com/news/imagen-4-shutdown-migration.html</guid>
      <pubDate>Fri, 14 Aug 2026 00:00:00 +0000</pubDate>
      <description>Google lists an August 17, 2026 shutdown date for three Imagen 4 endpoints. Migrate to Gemini 3.1 Flash Image by replacing the request and response adapter, then prove output behavior before cutover.</description>
    </item>
    <item>
      <title>Caching AI Responses Without Serving the Wrong Answer: Keys, TTLs, and Tenant Isolation</title>
      <link>https://aiprogrammingmanual.com/articles/caching-ai-responses.html</link>
      <guid>https://aiprogrammingmanual.com/articles/caching-ai-responses.html</guid>
      <pubDate>Thu, 13 Aug 2026 00:00:00 +0000</pubDate>
      <description>Design your own AI response cache for correctness: what belongs in the cache key, why TTL is the wrong invalidation for derived answers, how to handle errors and streams, and a runnable cross-tenant isolation test.</description>
    </item>
    <item>
      <title>Agent Trace Observability: What to Log So a Failed AI Agent Run Is Debuggable</title>
      <link>https://aiprogrammingmanual.com/articles/agent-trace-observability.html</link>
      <guid>https://aiprogrammingmanual.com/articles/agent-trace-observability.html</guid>
      <pubDate>Wed, 12 Aug 2026 00:00:00 +0000</pubDate>
      <description>Instrument AI agent runs with operation spans, trace propagation, provider request ids, cache-aware token accounting and opt-in content capture.</description>
    </item>
    <item>
      <title>Secure MCP Server Authorization: Prevent Confused-Deputy Tool Calls</title>
      <link>https://aiprogrammingmanual.com/articles/secure-mcp-server-authorization.html</link>
      <guid>https://aiprogrammingmanual.com/articles/secure-mcp-server-authorization.html</guid>
      <pubDate>Tue, 11 Aug 2026 00:00:00 +0000</pubDate>
      <description>Secure an HTTP MCP server with resource-bound tokens, audience validation, separate upstream credentials, per-client consent, and failure tests.</description>
    </item>
    <item>
      <title>Idempotent Tool Calls for AI Agents: Prevent Duplicate Side Effects</title>
      <link>https://aiprogrammingmanual.com/articles/idempotent-tool-calls-ai-agents.html</link>
      <guid>https://aiprogrammingmanual.com/articles/idempotent-tool-calls-ai-agents.html</guid>
      <pubDate>Mon, 10 Aug 2026 00:00:00 +0000</pubDate>
      <description>Design idempotent AI-agent tool calls that stay safe under timeouts, retries, and duplicate delivery, with a persistent operation record and failure tests.</description>
    </item>
    <item>
      <title>Zero Trust Architecture for Autonomous AI Agents</title>
      <link>https://aiprogrammingmanual.com/articles/zero-trust-architecture-ai-agents.html</link>
      <guid>https://aiprogrammingmanual.com/articles/zero-trust-architecture-ai-agents.html</guid>
      <pubDate>Sun, 09 Aug 2026 00:00:00 +0000</pubDate>
      <description>Allowing AI agents autonomous access to production data is a security risk without strict isolation.</description>
    </item>
    <item>
      <title>Setting Up Local AI Models for CI/CD Pipelines: Cost and Privacy Benefits</title>
      <link>https://aiprogrammingmanual.com/articles/local-ai-models-cicd-pipelines.html</link>
      <guid>https://aiprogrammingmanual.com/articles/local-ai-models-cicd-pipelines.html</guid>
      <pubDate>Fri, 07 Aug 2026 00:00:00 +0000</pubDate>
      <description>Relying on cloud APIs for AI support in CI/CD is expensive and leaks code. Learn how to run local models in your pipelines for testing and static analysis.</description>
    </item>
    <item>
      <title>Modeling AI Infrastructure Costs: Tokens, Storage, and Observability</title>
      <link>https://aiprogrammingmanual.com/articles/modeling-ai-infrastructure-costs.html</link>
      <guid>https://aiprogrammingmanual.com/articles/modeling-ai-infrastructure-costs.html</guid>
      <pubDate>Fri, 07 Aug 2026 00:00:00 +0000</pubDate>
      <description>Build a technical cost model that includes all infra components of an AI feature, not just the LLM calls.</description>
    </item>
    <item>
      <title>The AI Feature Flag Rollout Plan: Ship AI-Written Changes Behind a Toggle</title>
      <link>https://aiprogrammingmanual.com/articles/the-ai-feature-flag-rollout-plan.html</link>
      <guid>https://aiprogrammingmanual.com/articles/the-ai-feature-flag-rollout-plan.html</guid>
      <pubDate>Thu, 06 Aug 2026 00:00:00 +0000</pubDate>
      <description>Bind AI-written changes to feature flags: percentage rollout, a fast kill path, flag hygiene, and cleanup so flags never become permanent technical debt.</description>
    </item>
    <item>
      <title>AI Agent Rollback Plans: Approval Gates That Keep Production Recoverable</title>
      <link>https://aiprogrammingmanual.com/articles/ai-agent-rollback-approval-gates.html</link>
      <guid>https://aiprogrammingmanual.com/articles/ai-agent-rollback-approval-gates.html</guid>
      <pubDate>Wed, 05 Aug 2026 00:00:00 +0000</pubDate>
      <description>A practical way to approve AI agent actions in production: define the action boundary, require evidence, set rollback rules, and stop risky work before it becomes irreversible.</description>
    </item>
    <item>
      <title>How This Site Was Built: A Developer and AI, Start to Finish</title>
      <link>https://aiprogrammingmanual.com/articles/how-this-site-was-built.html</link>
      <guid>https://aiprogrammingmanual.com/articles/how-this-site-was-built.html</guid>
      <pubDate>Mon, 03 Aug 2026 00:00:00 +0000</pubDate>
      <description>The full story of building aiprogrammingmanual.com — one developer and Claude, over 6 weeks. What worked, what went wrong.</description>
    </item>
    <item>
      <title>Move a Synchronous Endpoint to a Queue: An AI-Assisted Migration Plan</title>
      <link>https://aiprogrammingmanual.com/articles/the-ai-queue-background-job-migration-plan.html</link>
      <guid>https://aiprogrammingmanual.com/articles/the-ai-queue-background-job-migration-plan.html</guid>
      <pubDate>Sun, 02 Aug 2026 00:00:00 +0000</pubDate>
      <description>Extract long-running work from an HTTP request into a new queue with explicit worker contracts, idempotency, backpressure, polling, and rollback.</description>
    </item>
    <item>
      <title>The AI Performance Fix Plan: Measure Before You Let AI Optimize</title>
      <link>https://aiprogrammingmanual.com/articles/the-ai-performance-fix-plan.html</link>
      <guid>https://aiprogrammingmanual.com/articles/the-ai-performance-fix-plan.html</guid>
      <pubDate>Fri, 31 Jul 2026 00:00:00 +0000</pubDate>
      <description>A safer AI-assisted workflow for performance work: baseline on realistic data, locate the bottleneck with a profiler or explain plan, set a target and a price, change one thing, and re-measure identically.</description>
    </item>
    <item>
      <title>AI Dev News</title>
      <link>https://aiprogrammingmanual.com/news/</link>
      <guid>https://aiprogrammingmanual.com/news/</guid>
      <pubDate>Mon, 27 Jul 2026 00:00:00 +0000</pubDate>
      <description>What changed in the AI stack — filtered for developers and vibe coders. Model releases, API changes, tool updates, and what they mean for how you build.</description>
    </item>
    <item>
      <title>Claude Opus 5 Keeps the Price and Changes the Bill</title>
      <link>https://aiprogrammingmanual.com/news/claude-opus-5-effort-dial.html</link>
      <guid>https://aiprogrammingmanual.com/news/claude-opus-5-effort-dial.html</guid>
      <pubDate>Mon, 27 Jul 2026 00:00:00 +0000</pubDate>
      <description>Claude Opus 5 shipped July 24, 2026 at Opus 4.8&apos;s exact list price. Thinking is now on by default, the five-level effort parameter is the main cost control, and the prompt-caching rules moved in two directions at once.</description>
    </item>
    <item>
      <title>The AI File Storage Migration Plan: Move Uploads Without Breaking Links</title>
      <link>https://aiprogrammingmanual.com/articles/the-ai-file-storage-migration-plan.html</link>
      <guid>https://aiprogrammingmanual.com/articles/the-ai-file-storage-migration-plan.html</guid>
      <pubDate>Sun, 19 Jul 2026 00:00:00 +0000</pubDate>
      <description>A safer AI-assisted workflow for moving uploads: inventory the storage contract, copy with checksum evidence, preserve access rules, reconcile live changes, and retire the old path safely.</description>
    </item>
  </channel>
</rss>
