<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Agent workflows | SpecDD</title><link>https://specdd.ai/how-to/agent-workflows/</link><description>Spec-driven development agent workflow guides for prompting, planning, reviewing, managing context, limiting scope, and keeping different agents aligned with specs.</description><generator>Hugo</generator><language>en-us</language><atom:link href="https://specdd.ai/how-to/agent-workflows/index.xml" rel="self" type="application/rss+xml"/><item><title>How to prompt an agent with SpecDD</title><link>https://specdd.ai/how-to/agent-workflows/how-to-prompt-an-agent-with-specdd/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-prompt-an-agent-with-specdd/</guid><description>Prompt an agent with spec-driven development by naming the spec, feature, task, or behavior while durable project context stays in local .sdd files.</description></item><item><title>How to make an agent follow specs</title><link>https://specdd.ai/how-to/agent-workflows/how-to-make-an-agent-follow-specs/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-make-an-agent-follow-specs/</guid><description>Make an agent follow SpecDD specs in a spec-driven development workflow by verifying setup, opening the right root, explaining the target spec, planning under authority, and reviewing the diff.</description></item><item><title>How to reduce agent mistakes with specs</title><link>https://specdd.ai/how-to/agent-workflows/how-to-reduce-agent-mistakes-with-specs/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-reduce-agent-mistakes-with-specs/</guid><description>Reduce agent mistakes with spec-driven development by making ownership, requirements, forbidden behavior, tasks, and verification explicit before implementation.</description></item><item><title>How to ask an agent to plan before coding</title><link>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-plan-before-coding/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-plan-before-coding/</guid><description>Ask an agent to plan before coding by using SpecDD specs in a spec-driven development workflow to identify authority, constraints, files, checks, and unresolved decisions.</description></item><item><title>How to review an agent plan against a spec</title><link>https://specdd.ai/how-to/agent-workflows/how-to-review-an-agent-plan-against-a-spec/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-review-an-agent-plan-against-a-spec/</guid><description>Review an agent plan against a SpecDD spec in a spec-driven development workflow by checking authority, required behavior, prohibited behavior, verification, and unresolved decisions.</description></item><item><title>How to keep agents from changing the wrong files</title><link>https://specdd.ai/how-to/agent-workflows/how-to-keep-agents-from-changing-the-wrong-files/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-keep-agents-from-changing-the-wrong-files/</guid><description>Keep agents from changing the wrong files by using spec-driven development Owns, Can modify, Can read, References, Must not, and Forbids correctly.</description></item><item><title>How to stop agents from overengineering</title><link>https://specdd.ai/how-to/agent-workflows/how-to-stop-agents-from-overengineering/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-stop-agents-from-overengineering/</guid><description>Stop agents from overengineering by using spec-driven development tasks, Done when criteria, Must not boundaries, and focused review before implementation expands.</description></item><item><title>How to use specs as agent memory</title><link>https://specdd.ai/how-to/agent-workflows/how-to-use-specs-as-agent-memory/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-use-specs-as-agent-memory/</guid><description>Use SpecDD specs as durable agent memory in a spec-driven development workflow by storing intent, boundaries, tasks, and completion criteria in local .sdd files instead of chat.</description></item><item><title>How to manage context for agents</title><link>https://specdd.ai/how-to/agent-workflows/how-to-manage-context-for-agents/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-manage-context-for-agents/</guid><description>Manage agent context with spec-driven development by choosing the right content root, relying on inherited specs, using references deliberately, and keeping prompts focused.</description></item><item><title>How to use SpecDD with multiple agents on one repo</title><link>https://specdd.ai/how-to/agent-workflows/how-to-use-specdd-with-multiple-agents-on-one-repo/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-use-specdd-with-multiple-agents-on-one-repo/</guid><description>Use spec-driven development with multiple agents on one repository by keeping specs in Git, assigning bounded tasks, preserving local authority, and reviewing task status.</description></item><item><title>How to get consistent results across different agents</title><link>https://specdd.ai/how-to/agent-workflows/how-to-get-consistent-results-across-different-agents/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-get-consistent-results-across-different-agents/</guid><description>Get consistent results across different agents with spec-driven development by using the same specs, content root, setup verification, prompts, plans, and review checks.</description></item><item><title>How to ask an agent to explain a spec before implementation</title><link>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-explain-a-spec-before-implementation/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-explain-a-spec-before-implementation/</guid><description>Ask an agent to explain a SpecDD spec before implementation in a spec-driven development workflow so you can confirm purpose, authority, constraints, tasks, and checks before coding.</description></item><item><title>How to stop an agent when the spec is ambiguous</title><link>https://specdd.ai/how-to/agent-workflows/how-to-stop-an-agent-when-the-spec-is-ambiguous/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-stop-an-agent-when-the-spec-is-ambiguous/</guid><description>Stop an agent when a SpecDD spec is ambiguous in a spec-driven development workflow by identifying unresolved decisions, marking tasks correctly, updating the spec, and resuming only after review.</description></item><item><title>How to ask an agent to update the spec after discovering missing context</title><link>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-update-the-spec-after-discovering-missing-context/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-ask-an-agent-to-update-the-spec-after-discovering-missing-context/</guid><description>Ask an agent to update a SpecDD spec in a spec-driven development workflow after missing context is discovered by proposing, reviewing, applying, and verifying a contract change before coding.</description></item><item><title>How to keep agent prompts short with SpecDD</title><link>https://specdd.ai/how-to/agent-workflows/how-to-keep-agent-prompts-short-with-specdd/</link><guid>https://specdd.ai/how-to/agent-workflows/how-to-keep-agent-prompts-short-with-specdd/</guid><description>Keep agent prompts short with spec-driven development by moving durable context into specs, naming the task, using one instruction, and reviewing output against the contract.</description></item></channel></rss>