IdeaWell

Project ideas from Hacker News discussions.

Credit where credit is due

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Three prevalent themes in the discussion

  1. Accountability — the person who signs off is responsible, regardless of how the work was produced
  2. ChrisMarshallNY: “If I sign off on something; no matter how it was made, it's my ass on the line. If I trust a shyster, and sign off on junk, then I walk the plank, not the shyster.”

  3. pixl97: “People no longer understand Accountability … A fish rots from the head. The increase in recording and reporting has shown us the people at the top are rarely held accountable and make billions doing so.”

  4. LLMs as just another tool — their use is not inherently cheating

  5. bassie123: “I wonder what it was like when we first got calculators; people saying its cheating at arithmetic. AI is just the next tool humans can apply to problems to solve.”

  6. clickety_clack: “The ‘real public’ wants math to advance so that we can reap the benefits. We don’t care what tools you are or are not using, grant money will go to whichever group is making more progress.”

  7. Transparency/authorship — whether (and how) to credit LLM assistance

  8. pxc: “I use LLMs every day at work. At the same time, I've never had an LLM appear as a 'co-author', because 'adding an LLM as co-author gives you a way for you to shirk responsibility for mistakes.'”
  9. elliotmr: “If I write something and ask an LLM to edit or critique that piece, is it LLM generated or is it mine? … I usually follow this procedure for formal emails … I am just trying to convey my meaning accurately … I believe that I get more precise results with an LLM helper.”

🚀 Project Ideas

LLM Accountability Gate

Summary

  • A git hook and CI integration that prevents commits containing LLM‑generated code from being merged without explicit human sign‑off, addressing the accountability gap when LLMs contribute code.
  • Core value: guarantees that every pushed change carries a human attestation, preserving personal brand and trust.

Details

Key Value
Target Audience Software engineers, teams using LLMs for coding assistance
Core Feature Detects LLM‑authored diff (via markers or model fingerprint) and requires a signed‑off‑by trailer or approval before allowing push/merge
Tech Stack Git hooks (bash), GitHub Actions/GitLab CI, optional Rust detector, JSON log store
Difficulty Medium
Monetization Revenue-ready: SaaS tier for hosted detection & audit dashboard ($10/user/mo)

Notes

  • HN commenters stressed “If I sign off on something; no matter how it was made, it’s my ass on the line” (ChrisMarshallNY) and “I don't want LLMs performing any git operations that add or push commits” (pxc). This tool enforces that rule.
  • Enables discussion on responsible AI use and provides auditable logs for compliance or code‑review culture.

CodeProvenance Service

Summary

  • A lightweight attestation service that signs each commit with metadata indicating whether the author is human, LLM‑assisted, or LLM‑generated, creating an immutable provenance trail.
  • Core value: lets teams verify and showcase accountability, satisfying the desire for transparency and integrity in code contributions.

Details

Key Value
Target Audience Open‑source maintainers, enterprise dev teams, auditors
Core Feature Generates Sigstore/GPG signatures attached to commits with LLM‑usage flags; provides verification UI and API
Tech Stack Sigstore (cosign), Go backend, PostgreSQL for metadata, Web dashboard (React)
Difficulty High
Monetization Revenue-ready: Subscription for private repos ($20/mo per org) + free public tier

Notes

  • Commenters lamented loss of accountability and wanted to “stand behind their work” (ChrisMarshallNY); CodeProvenance gives a cryptographic way to do that.
  • Provides concrete data for discussions about AI‑generated code and can be integrated into CI to block unsigned commits.

LLM-Assist Review Assistant (IDE Plugin)

Summary

  • An IDE extension that surfaces LLM suggestions as review comments or inline diffs, never writes code directly, forcing the developer to accept or reject each change manually.
  • Core value: preserves human agency and accountability while still gaining LLM productivity, matching the workflow many developers already use (pxc, elliotmr).

Details

Key Value
Target Audience Individual developers using VS Code, JetBrains, or NeoVim
Core Feature Shows LLM‑generated edits as reviewable suggestions; logs all interactions for accountability audit
Tech Stack Language Server Protocol extension, TypeScript/VS Code API, optional local LLM via Ollama, telemetry opt‑out
Difficulty Low
Monetization Hobby (open‑source) – can accept donations or sponsorship

Notes

  • pxc said “I don't want LLMs performing any git operations that add or push commits”; this plugin respects that by keeping edits in the editor only.
  • elliotmr described using LLM for review, not generation; this tool formalizes that pattern and provides a traceable log for accountability discussions.

Read Later