{"id":46384407,"url":"https://github.com/atakanatali/agent-behavior-development","last_synced_at":"2026-03-05T07:05:04.149Z","repository":{"id":333789120,"uuid":"1138691711","full_name":"atakanatali/agent-behavior-development","owner":"atakanatali","description":"ABD: A development paradigm where the primary refactoring target is agent behavior (prompts, guardrails, flow), and code is a byproduct validated by evidence.","archived":false,"fork":false,"pushed_at":"2026-02-14T09:42:37.000Z","size":634,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-14T18:03:25.511Z","etag":null,"topics":["agent-behavior-development","agile-methodology","continuous-improvement","human-in-the-loop","prompt-as-artifact","prompt-engineering","scrum-compatible","software-development-paradigm","software-quality"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/atakanatali.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":"GOVERNANCE.md","roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-01-21T02:00:45.000Z","updated_at":"2026-02-14T09:42:38.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/atakanatali/agent-behavior-development","commit_stats":null,"previous_names":["atakanatali/agent-behavior-development"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/atakanatali/agent-behavior-development","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/atakanatali%2Fagent-behavior-development","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/atakanatali%2Fagent-behavior-development/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/atakanatali%2Fagent-behavior-development/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/atakanatali%2Fagent-behavior-development/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/atakanatali","download_url":"https://codeload.github.com/atakanatali/agent-behavior-development/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/atakanatali%2Fagent-behavior-development/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30113130,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-05T03:40:26.266Z","status":"ssl_error","status_checked_at":"2026-03-05T03:39:15.902Z","response_time":93,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agent-behavior-development","agile-methodology","continuous-improvement","human-in-the-loop","prompt-as-artifact","prompt-engineering","scrum-compatible","software-development-paradigm","software-quality"],"created_at":"2026-03-05T07:05:02.274Z","updated_at":"2026-03-05T07:05:04.138Z","avatar_url":"https://github.com/atakanatali.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Agent Behavior Development (ABD)\n\nAgent Behavior Development (ABD) is a software development paradigm that treats\n**agent behavior** as the primary engineering artifact.\n\nIn ABD:\n- Code is not the first-class citizen.\n- Prompts, guardrails, output formats, review rules, and recycle loops are.\n\nThis repository does not provide tools.\nIt defines a **development paradigm**, not an SDK.\n\n---\n\n## One-Sentence Definition\n\n\u003e **Agent Behavior Development (ABD)** is a paradigm where humans design and evolve the behavior of agents, and code is a byproduct validated by evidence.\n\n---\n\n```mermaid\ngraph TB\n\n  subgraph Planning\n    A[Product Intent or Sprint Goal]\n    B[Notlanot Sprint Plan with Personas]\n    C[Task Packets]\n    D[Conflict Inputs]\n    E[Conflict Scoring and Replan]\n  end\n\n  subgraph PromptOps\n    F[Prompt Library]\n    G[Prompt Compiler]\n    H[Guardrails]\n    I[Output Format Contract]\n  end\n\n  subgraph Execution\n    J[Agent Run]\n    K[Structured Output]\n    L[Evidence Production]\n  end\n\n  subgraph ReviewRecycle\n    M[Scorecard]\n    N[Tags and Anti Patterns]\n    O[Prompt Patch Rules]\n    P[Promote to Library]\n    Q[Recycle Outputs]\n    R[Changelog and Governance]\n  end\n\n  A --\u003e B\n  B --\u003e C\n  C --\u003e D\n  D --\u003e E\n\n  E --\u003e G\n  F --\u003e G\n  G --\u003e H\n  H --\u003e I\n  I --\u003e J\n\n  J --\u003e K\n  K --\u003e L\n  K --\u003e M\n  L --\u003e M\n\n  M --\u003e N\n  N --\u003e O\n  O --\u003e J\n  O --\u003e P\n  P --\u003e F\n\n  M --\u003e Q\n  Q --\u003e R\n  R --\u003e B\n\n  style F fill:#9C27B0,color:#fff\n  style M fill:#FF9800,color:#fff\n  style O fill:#F44336,color:#fff\n  style L fill:#4CAF50,color:#fff\n  style I fill:#1565C0,color:#fff\n```\n\n---\n\n## Why ABD Exists\n\nTraditional development paradigms assume:\n- Humans write specifications.\n- Humans write tests.\n- Humans write most of the code.\n- Tools assist.\n\nThat assumption no longer holds.\n\nModern agents:\n- Propose behaviors\n- Generate test strategies\n- Produce implementation drafts\n\nYet teams still:\n- Debug code instead of behavior\n- Retry prompts instead of fixing agent constraints\n- Treat prompt failures as one-off accidents\n\nABD exists to systematize agent behavior improvement.\n\n---\n\n## The Core Shift: Code → Behavior\n\nThe defining shift of ABD is simple:\n\n\u003e **The primary refactoring target is agent behavior, not code.**\n\nIn ABD, when output quality is poor:\n1) Fix the prompt\n2) Tighten guardrails\n3) Lock output formats\n4) Improve review rules\n5) Strengthen recycle logic  \n**Only then** adjust code.\n\n---\n\n## What ABD Is\n\nABD is:\n- Agent-first, not agent-autonomous\n- Behavior-driven, not prompt-driven\n- Evidence-oriented, not test-obsessed\n- Governance-heavy, not “just try again”\n\nABD introduces:\n- Prompt-as-artifact discipline\n- Behavior scorecards\n- Mandatory recycle outputs\n- Conflict-aware task planning\n- Agent behavior retrospectives\n\n---\n\n## What ABD Is Not\n\nABD is NOT:\n- AI-assisted TDD\n- BDD automation\n- “Let the agent do everything”\n- Prompt engineering tips\n- A replacement for Scrum\n\nABD complements Scrum by adding an **agent behavior governance layer**.\n\n---\n\n## ABD vs TDD vs BDD\n\n| Aspect | TDD | BDD | ABD |\n|-----|-----|-----|-----|\n| Who defines behavior | Human | Human | Agent (under rules) |\n| Primary artifact | Test | Scenario | Prompt + Guardrails |\n| What is refactored | Code | Code | Agent behavior |\n| Failure resolution | Rewrite code | Rewrite scenario | Patch prompt/flow |\n| Evidence | Tests | Scenarios | Tests, checks, demos |\n| Paradigm focus | Code correctness | Business behavior | Behavior production system |\n\nTDD and BDD are **techniques** in ABD, not paradigms.\n\n---\n\n## Non-Negotiable Rules (v1)\n\n1) Agent output must follow a fixed format.\n2) Evidence must be proposed before implementation.\n3) Assumptions must be explicit.\n4) Agents may ask up to 3 questions and must stop.\n5) Every task must produce recycle outputs.\n6) Behavior changes must be versioned.\n\n---\n\n## Repository Map\n\n- `MANIFESTO.md` → Hard rules of the paradigm\n- `0_introduction/` → What ABD is and is not\n- `1_core/` → Principles and artifacts\n- `2_process/` → Lifecycle, sprint and task models\n- `3_promptops/` → Prompt operations and metrics\n- `4_templates/` → Ready-to-use templates\n- `5_examples/` → End-to-end worked examples\n\n---\n\n## First-Week Adoption Guide\n\nDay 1:\n- Read MANIFESTO\n- Use the core prompt template for one task\n- Score the output\n\nDay 2–3:\n- Run 3 more tasks\n- Tag recurring failures\n\nDay 4:\n- Patch the prompt\n- Document behavior improvement\n\nDay 5:\n- Sprint review focused on agent behavior, not features\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fatakanatali%2Fagent-behavior-development","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fatakanatali%2Fagent-behavior-development","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fatakanatali%2Fagent-behavior-development/lists"}