NASA-IMPACT / NASA-IMPACT/akd-core

AKD Code Search Agent preprint

Open
#486 1 comment 0 reactions 1 assignee View on GitHub

@pushwithak is already working on this.

Since Jun 29, 2026.

Dominant language
Jupyter Notebook
Stars
4
Forks
0
PR merge metrics
No merged PRs in 30d

Description

Preprint: AKD Code Search Agent
Paper repo: NASA-IMPACT/code-search-paper-draft-v1"Datasets & Benchmark paper for code search"
Parent EPIC: #474 — Preprint Papers
Tracking: Milestone 26.4 Paper and Workshop
Owner: @pushwithak
Deadline: Tentative — mid-July 2026 (no hard deadline set).

Track the AKD Code Search Agent preprint from draft to submission-ready. This is framed as a datasets & benchmark paper, so the benchmarking/evaluation section carries extra weight. Tick the checklist as items land; link evidence (PRs, figures, eval runs, draft sections) inline.


📋 Paper Readiness Checklist

Literature & positioning
  • Literature review complete — related work surveyed, gaps identified, key prior art cited
  • Contribution / novelty stated explicitly relative to prior work (and vs. existing code-search benchmarks)
Agent architecture & methodology
  • Architecture diagram(s) for the agent(s) included — components, tools, data flow, control loop
  • Methodology described in enough detail to reproduce
  • Design decisions and trade-offs justified
Benchmarking & evaluation
  • Benchmark dataset(s) defined, sourced, and documented — construction, size, splits, licensing
  • Evaluation is rigorous — clear metrics, baselines / comparisons, ablations where relevant
  • Results reproducible — code, configs, and data referenced (link akd-evals or the eval harness used)
  • Quantitative results reported with appropriate rigor — multiple runs / error bars / significance where applicable
Science narrative & impact
  • Science / developer use-case narrative written — which research-software workflow this accelerates and for whom
  • Impact articulated — value to NASA science software / the target research community
Limitations & responsible AI
  • Limitations section written — honest scope, known failure modes, where it breaks down
  • Responsible-AI / ethical considerations addressed (where applicable)
Manuscript & submission
  • All sections drafted — abstract, intro, methods, results, discussion, conclusion
  • Figures and tables finalized
  • References complete and consistently formatted
  • Internal review / proofread complete
  • Target venue / preprint server chosen (e.g. arXiv) and author list + contributions agreed

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.