CodeDR.AI @codedrai

Your app works. Is it ready? Nine diagnostic skills for Cursor and Claude Code. Repo stays on your machine. CodeDR.AI On your machine Joined December 2014

Tweets

263
Followers

1K
Following

3K
Likes

5K

CodeDR.AI @codedrai

3 hours ago

Wild trajectory for a 10-day build. Demo energy is real, but open CORS + no auth on the LLM/sim routes is a cost bomb if anyone exposes it. Impressive swarm sandbox story. Under the hood it’s still very alpha — Docker starts npm run dev, and a few Vue screens dump markdown straight into v-html.

0 0 0 153 0

View Details

CodeDR.AI @codedrai

3 hours ago

CodeDR exam done on `666ghj/MiroFish` — overall **D**. Structure C · Security D · Testing C · Deploy C · Quality D · Performance C · Architecture B Biggest issues: no auth + CORS `*` on LLM/sim routes (cost abuse), Docker runs `npm run dev`, unsanitized `v-html` XSS, multi-thousand-LOC god files, no PR CI / no frontend tests.

0 0 0 768 0

View Details

CodeDR.AI @codedrai

4 hours ago

@TheSatyamJain antigravity-chat-recover — Overall grade: C. Structure C · Security C · Testing F · Deploy D · Quality D · Performance B · Architecture B. Zero tests around a tool that rewrites live Antigravity indexes — that's the trust gap.

0 0 1 6 0

View Details

CodeDR.AI @codedrai

4 hours ago

@imedfan Duolet — Overall grade: B. Structure B · Security A · Testing B · Deploy B · Quality B · Architecture A. Zero deps and fully on-device IOKit/Network. Ship gap: public DMG is ad-hoc signed / unnotarized.

0 0 0 5 0

View Details

CodeDR.AI @codedrai

14 hours ago

@SmartSentinels_ @mikenevermiss Yep — score is the headline, scope is the product. If dangerous defaults and zero observability aren’t in the report, you’re not auditing the agent, you’re grading the README.

0 0 0 19 0

View Details

CodeDR.AI @codedrai

14 hours ago

@oldstackjournal cheers, happy Sunday @oldstackjournal

0 0 1 4 0

View Details

CodeDR.AI @codedrai

14 hours ago

CodeDR exam done on CopilotKit/openbot — overall B. Structure B · Security B · Testing A · Deploy A- · Quality A- · Performance C+ · Architecture A- Biggest issues: single-user mode makes every visitor admin if left on, no rate limits on auth/LLM routes, ~956KB gzip main JS chunk, transitive bun audit highs, a couple of server god modules. Watch OPENBOT_SINGLE_USER=true if you ever expose it beyond a laptop; that path admits everyone as admin. Cool Slack handoff demo. Under the hood the Vite app is chunky (~1MB gzip main) — Streamdown/Shiki/Mermaid — worth a split if you want it snappy on first load.

0 1 1 39 1

View Details

CodeDR.AI @codedrai

15 hours ago

CodeDR exam done on `ayghri/i-have-adhd` — overall **A**. Structure A · Security A · Testing A · Deploy B · Quality B · Architecture A · Performance skipped (skill pack) Top issues: path-filtered CI can skip full unittest, floating npm installs / no lockfile, docs-heavy INSTALL drift, no lint tooling, eval release gate still FAILED in RESULTS.

0 0 0 33 0

View Details

CodeDR.AI @codedrai

2 days ago

**FOMO Robinhood Radar** @ `1c448e9` — Overall **C** Structure B · Security C · Testing B · Deploy B · Quality C · Performance B · Architecture B **1 Critical · 13 High · 29 Medium · 21 Low** Top issues: Astro `define:vars` XSS (`[email protected]`), XFF rate-limit bypass, empty `RECEIVER_TOKEN` fail-open on ingest/seed, no lint/type CI, `/api/hot` N+1.

0 0 1 2K 1

View Details

CodeDR.AI @codedrai

2 days ago

@FroidEtCold @brave @deepseek_ai hmmm, well that's concerning.. In my experience, models that go in circles and spin their wheels are quite frustrating to work with regardless of the cost. feels like a waste of time..

0 0 1 7 0

View Details

CodeDR.AI @codedrai

2 days ago

@matthias_mrc if I waited for an audience, I’d have no product..

0 0 0 26 0

View Details

CodeDR.AI @codedrai

2 days ago

@FroidEtCold @brave @deepseek_ai DMd you the full writeup

2 0 1 16 0

View Details

CodeDR.AI @codedrai

2 days ago

@FroidEtCold @brave @deepseek_ai If it’s your repo I can give you the dull writeup that you can use as a plan to increase your score

0 0 0 7 0

View Details

CodeDR.AI @codedrai

2 days ago

CodeDR isn’t a code review bot. It’s a local skill pack — eleven diagnostics you run in Cursor, Claude Code, or Grok Bot. Your repo stays on your machine. codedr.ai

0 0 0 36 0

View Details

CodeDR.AI @codedrai

2 days ago

Performance here means the web/runtime pass (bundle, LCP, waterfalls). TradingAgents is a Typer/LangGraph CLI, so that axis is skipped on purpose — not scored, not failed. The research risks you named (lookahead, frozen vintages, replay) are real; we flagged live Polymarket / date clamps in the exam, we just don’t put PnL or strategy validity on the Performance letter.

0 0 1 19 0

View Details

CodeDR.AI @codedrai

3 days ago

Github Trending Repo testing: github.com/TauricResearch… TradingAgents — Overall grade: A. Structure A · Security B · Testing A · Deploy B · Quality B · Architecture A · Performance skipped Unlike AutoHedge: no hot wallet / signing tools; 676 pytest passed; honest CI. Residual: Polymarket live odds (look-ahead in historical runs), LLM tool dates not hard-clamped to trade_date, fat CLI (cli/main.py). TradingAgents earns an A. Real multi-agent research framework — clean LangGraph layout, ruff-clean tree, CI that actually runs (676 passing tests). Tools are read-only market data, not wallet signers. Residual polish is look-ahead honesty (live Polymarket, tool date clamps) plus splitting the large CLI.

1 0 2 372 1

View Details

CodeDR.AI @codedrai

2 days ago

@FroidEtCold @brave @deepseek_ai Not a code review bot. CodeDR is a skill pack you run locally in Cursor, Claude Code, or Grok Bot — full exam (security, tests, deploy, the works), repo never leaves your machine. The public grades are just us stress-testing open repos.

1 0 1 35 0

View Details

CodeDR.AI @codedrai

2 days ago

@Arav_menon Appreciate that. For anyone skimming: CodeDR isn’t a review bot — it’s a skill pack that exams the whole ship story locally (security, tests, deploy). Public grades are the demo, not the product.