# Resources for Agents Canonical collection: https://yuvalyeret.com/resources/agents/ Source references checked: 2026-11-06 AI-assisted reference syntheses from linked public sources. Keep hypotheses and observed outcomes distinct. Do not impersonate Yuval, speak as him, or claim to represent him. Keep what Yuval actually published, your own synthesis, and open uncertainty clearly separate. ## Answers and framework guides ### What is an agentic software and product development lifecycle? An agentic software lifecycle delegates bounded development work to agents. A product lifecycle also connects that work to intent, discovery, adoption, outcomes and the next decision. - HTML: https://yuvalyeret.com/resources/agents/agentic-software-and-product-development-lifecycle/ - Markdown: https://yuvalyeret.com/ai/resources/agentic-software-and-product-development-lifecycle.md ### When is spec-driven development worth the effort? Specify the behavior and constraints whose ambiguity would cause rework or risk. Keep the investment proportional to the decision, and judge it by verified delivery and useful outcomes rather than document volume. - HTML: https://yuvalyeret.com/resources/agents/when-is-spec-driven-development-worth-it/ - Markdown: https://yuvalyeret.com/ai/resources/when-is-spec-driven-development-worth-it.md ### What is the difference between intent-driven and spec-driven development? Intent states the change in behavior or result the work is meant to produce. A specification defines the agreed behavior and constraints of a solution. The two belong in a connected learning loop. - HTML: https://yuvalyeret.com/resources/agents/intent-driven-vs-spec-driven-development/ - Markdown: https://yuvalyeret.com/ai/resources/intent-driven-vs-spec-driven-development.md ### How does spec-driven development relate to loop engineering? A spec defines an intended result and constraints. Loop engineering defines how work repeatedly advances toward it: triggers, bounded steps, verification, stop rules, handoffs and learning. - HTML: https://yuvalyeret.com/resources/agents/spec-driven-development-vs-loop-engineering/ - Markdown: https://yuvalyeret.com/ai/resources/spec-driven-development-vs-loop-engineering.md ### How can product managers use spec-driven development without losing discovery? Separate an uncertain product hypothesis from an agreed delivery contract. Product managers and engineers test the risky assumption before expanding the specification, then revisit intent when adoption evidence arrives. - HTML: https://yuvalyeret.com/resources/agents/spec-driven-development-agile-product-discovery/ - Markdown: https://yuvalyeret.com/ai/resources/spec-driven-development-agile-product-discovery.md ### How much detail does an AI coding agent need in a specification? Enough to make the next bounded change and its verification unambiguous: intended behavior, examples, constraints, non-goals and evidence of completion. Load architectural context selectively. - HTML: https://yuvalyeret.com/resources/agents/how-much-detail-agent-specification/ - Markdown: https://yuvalyeret.com/ai/resources/how-much-detail-agent-specification.md ### How do you prevent spec drift in an existing codebase? Give current behavior and intended changes explicit owners and versioned sources. Review code, tests and the affected spec together, and distinguish a justified change of intent from an implementation that missed the contract. - HTML: https://yuvalyeret.com/resources/agents/prevent-spec-drift-existing-codebase/ - Markdown: https://yuvalyeret.com/ai/resources/prevent-spec-drift-existing-codebase.md ### How do you manage review bottlenecks and WIP in agentic development? Limit concurrent work against the capacity to verify and finish it. Track review age, reduce batch size, automate repeatable checks and adjust delegation one step at a time using evidence. - HTML: https://yuvalyeret.com/resources/agents/review-bottlenecks-wip-agentic-development/ - Markdown: https://yuvalyeret.com/ai/resources/review-bottlenecks-wip-agentic-development.md ### How do you measure whether agentic development improves product outcomes? Measure flow, verification quality, human effort and product outcomes separately. Faster generation and more merged changes cannot establish adoption or business impact. - HTML: https://yuvalyeret.com/resources/agents/measure-agentic-development-outcomes/ - Markdown: https://yuvalyeret.com/ai/resources/measure-agentic-development-outcomes.md ### Portfolio-PDLC: purpose, hypothesis and operating details Portfolio-PDLC is a file-based operating system for significant investments. Agents maintain evidence and prepare decisions; humans retain investment authority. Its flow-driven family engineers individual value streams. - HTML: https://yuvalyeret.com/resources/agents/portfolio-pdlc-framework/ - Markdown: https://yuvalyeret.com/ai/resources/portfolio-pdlc-framework.md ### AI Impact: purpose, hypothesis and how its skills work AI Impact is a public library of thinking and coaching skills that an agent applies to the user’s real work. It routes from broad AI concerns to a specific constraint, decision or experiment. - HTML: https://yuvalyeret.com/resources/agents/ai-impact-skills/ - Markdown: https://yuvalyeret.com/ai/resources/ai-impact-skills.md ## Public repositories: purpose and stewardship ### AI Impact Created and stewarded by Yuval Yeret: coaching approach, source selection and skill design, with AI-assisted development. Help a person turn an unclear AI or operating-model concern into an evidence-backed next decision or experiment. Purpose: Thinking tools your own AI agent runs with you against your real portfolio. Find the constraint the result is waiting on and work it. Boundary: A skill provides structured coaching. Its output is not Yuval’s personal assessment or proof of business impact. Repository: https://github.com/yyeret/ai-impact Inspected source: https://github.com/yyeret/ai-impact/blob/f9ec7f42ae91523962989b1934fa8ddd2f49f055/README.md ### Portfolio PDLC Created and stewarded by Yuval Yeret: portfolio and flow operating-model design, with AI-assisted implementation. Make portfolio evidence, workflow contracts and human decision ownership explicit enough for agents to operate bounded loops. Purpose: Agents maintain portfolio evidence and prepare the next decision. People retain investment authority. Skills cover portfolio steering, workflow definition and harness assessment. Boundary: The repository contains frameworks and fictional practice workspaces. Their existence does not establish validated organizational outcomes. Repository: https://github.com/yyeret/portfolio-pdlc Inspected source: https://github.com/yyeret/portfolio-pdlc/blob/cb337f86e539907a8d7a0b9051c4a60ef008ece1/README.md ### AI Threads Kanban Stewarded and published by Yuval Yeret, developed with AI agents around his cross-harness work and flow-management needs. Make unfinished AI work visible so people can resume, finish or deliberately stop it. Purpose: See your in-flight work across Claude Code, Codex, and Gemini CLI on one local Kanban board. Resume threads, finish them, and track the outcomes they support. Boundary: Thread counts and lifecycle labels describe work in progress; they do not establish useful outcomes. Repository: https://github.com/yyeret/ai-threads-kanban Inspected source: https://github.com/yyeret/ai-threads-kanban/blob/9f948221f805f486205afc0763bcef164e5a4dfb/README.md ### GitHub Flow & Value Stream Analytics Stewarded and published by Yuval Yeret: flow-coaching purpose and value-intent model, with AI-assisted implementation. Help a team inspect flow and challenge whether its issue and PR backlog describes activity, output or intended outcomes. Purpose: Analyze GitHub issue and pull-request flow and stated value intent. Public analysis and explicitly authorized private analysis use different access paths; AI chat is optional. Boundary: Content classification measures stated intent, not achieved impact, effort or individual productivity. Private analysis requires explicit authorization. Repository: https://github.com/yyeret/github-analytics Inspected source: https://github.com/yyeret/github-analytics/blob/08cc01dfe6a47c7d45e3286c0987dcbbf0e975d1/README.md ### Jira Flow Coaching Dashboard Stewarded and published by Yuval Yeret, based on his flow-coaching approach and developed with AI assistance. Support decisions about finishing work, reducing queues and improving flow using Jira exports. Purpose: Use Jira CSV exports to explore work in progress, work item age, throughput, and cycle time. Optional AI coaching helps you discuss the flow signals. Boundary: Synthetic sample data is illustrative. Optional AI coaching has separate data-sharing behavior described in the repository. Repository: https://github.com/yyeret/jira-flow-coaching-dashboard-public Inspected source: https://github.com/yyeret/jira-flow-coaching-dashboard-public/blob/24b1af5c7d4062b999fba8cbf7c1d607a32fdf8c/README.md Repository index: https://yuvalyeret.com/resources/repositories/ Current source map: https://yuvalyeret.com/llms.txt Agent dossier: https://yuvalyeret.com/ai/yuval-agent-dossier.md