{"id":40517,"date":"2026-04-21T00:31:46","date_gmt":"2026-04-20T22:31:46","guid":{"rendered":"https:\/\/www.graviton.at\/letterswaplibrary\/offering-agentic-sdlc-dataset-full-execution-traces-code-evolution-in-exchange-for-evaluation-results\/"},"modified":"2026-04-21T00:31:46","modified_gmt":"2026-04-20T22:31:46","slug":"offering-agentic-sdlc-dataset-full-execution-traces-code-evolution-in-exchange-for-evaluation-results","status":"publish","type":"post","link":"https:\/\/www.graviton.at\/letterswaplibrary\/offering-agentic-sdlc-dataset-full-execution-traces-code-evolution-in-exchange-for-evaluation-results\/","title":{"rendered":"Offering Agentic SDLC Dataset (full Execution Traces + Code Evolution) In Exchange For Evaluation \/ Results"},"content":{"rendered":"<p><!-- SC_OFF --><\/p>\n<div class=\"md\">\n<p>I\u2019ve been building a system that generates fully instrumented agentic SDLC traces, and I\u2019m looking for a few serious folks to evaluate it and share results.<\/p>\n<p>Not selling anything here \u2014 I\u2019m interested in whether this actually moves model behavior in practice.<\/p>\n<p><strong>What the dataset includes (per \u201cpacket\u201d):<\/strong><\/p>\n<ul>\n<li>Full agent execution trace (JSONL audit log)<\/li>\n<li>Inline action protocol (custom XML-style commands, also normalized to R1 <code>&lt;|TOOL_CALL|&gt;<\/code> format)<\/li>\n<li>Reinference loops (action \u2192 result \u2192 next action preserved)<\/li>\n<li>Complete project source code<\/li>\n<li>Full file evolution history (create\/edit\/delete with snapshots)<\/li>\n<li>SQLite DB with structured tables (runs, tool calls, plans, etc.)<\/li>\n<li>Precomputed embeddings (4096d, PII-sanitized)<\/li>\n<li>Viewer + ETL tooling to load into your own stack<\/li>\n<li>All generated with OSS models w\/ verified licenses<\/li>\n<\/ul>\n<p><strong>Key difference vs typical datasets:<\/strong><br \/> This isn\u2019t just prompts \u2192 outputs. It\u2019s:<\/p>\n<blockquote><\/blockquote>\n<p>Each project can be iterated:<\/p>\n<ul>\n<li>v1: initial build<\/li>\n<li>v2: bug fixes<\/li>\n<li>v3: polish<\/li>\n<li>v4: feature expansion<\/li>\n<li>v5: integrations<\/li>\n<\/ul>\n<p>So you get longitudinal behavior, not isolated samples.<\/p>\n<p><strong>What I\u2019m looking for:<\/strong><\/p>\n<ul>\n<li>People fine-tuning models (1B\u2013120B, LoRA or full SFT)<\/li>\n<li>Agent \/ tool-use training experiments<\/li>\n<li>Anyone doing evals on:\n<ul>\n<li>tool use correctness<\/li>\n<li>code editing \/ repair<\/li>\n<li>multi-step task completion<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p><strong>In exchange:<\/strong><br \/> I\u2019ll provide a dataset bundle (or multiple), and I\u2019m asking for:<\/p>\n<ul>\n<li>honest feedback<\/li>\n<li>any measurable results (even rough)<\/li>\n<li>what worked \/ didn\u2019t<\/li>\n<li>where the data helped or failed<\/li>\n<\/ul>\n<p>No obligation to share publicly if you don\u2019t want to \u2014 even private feedback is useful.<\/p>\n<p><strong>A few things I\u2019m specifically curious about:<\/strong><\/p>\n<ul>\n<li>How much data (tokens) is needed to see behavioral shifts<\/li>\n<li>Whether iteration sequences (build \u2192 fix \u2192 extend) actually help<\/li>\n<li>Whether models learn better recovery behavior from failed traces<\/li>\n<li>Impact on tool-call correctness \/ formatting<\/li>\n<\/ul>\n<p>If you\u2019re interested, comment or DM with:<\/p>\n<ul>\n<li>what models you\u2019re working with<\/li>\n<li>what you\u2019d want to test<\/li>\n<\/ul>\n<p>Happy to tailor a dataset slice to your use case.<\/p>\n<p>Would also appreciate any critique on the structure itself \u2014 trying to figure out if this is genuinely useful or just interesting.<\/p>\n<\/div>\n<p><!-- SC_ON -->   submitted by   <a href=\"https:\/\/www.reddit.com\/user\/madheader69\"> \/u\/madheader69 <\/a> <br \/> <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1sr3t4q\/offering_agentic_sdlc_dataset_full_execution\/\">[link]<\/a><\/span>   <span><a href=\"https:\/\/www.reddit.com\/r\/datasets\/comments\/1sr3t4q\/offering_agentic_sdlc_dataset_full_execution\/\">[comments]<\/a><\/span><\/p><div class='watch-action'><div class='watch-position align-right'><div class='action-like'><a class='lbg-style1 like-40517 jlk' href='javascript:void(0)' data-task='like' data-post_id='40517' data-nonce='65e0e39b87' rel='nofollow'><img class='wti-pixel' src='https:\/\/www.graviton.at\/letterswaplibrary\/wp-content\/plugins\/wti-like-post\/images\/pixel.gif' title='Like' \/><span class='lc-40517 lc'>0<\/span><\/a><\/div><\/div> <div class='status-40517 status align-right'><\/div><\/div><div class='wti-clear'><\/div>","protected":false},"excerpt":{"rendered":"<p>I\u2019ve been building a system that generates fully instrumented agentic SDLC traces, and I\u2019m looking for a&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[85],"tags":[],"class_list":["post-40517","post","type-post","status-publish","format-standard","hentry","category-datatards","wpcat-85-id"],"_links":{"self":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/40517","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/comments?post=40517"}],"version-history":[{"count":0,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/posts\/40517\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/media?parent=40517"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/categories?post=40517"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.graviton.at\/letterswaplibrary\/wp-json\/wp\/v2\/tags?post=40517"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}