Comment by mrlesk

6 days ago

I threw Claude Code at an existing codebase a few months back and quickly quit— untangling its output was slower than writing from scratch. The fix turned out to be process, not model horsepower.

Iteration timeline

==================

• 50 % task success - added README.md + CLAUDE.md so the model knew the project.

• 75 % - wrote one markdown file per task; Codex plans, Claude codes.

• 95 %+ - built Backlog.md, a CLI that turns a high-level spec into those task files automatically (yes, using Claude/Codex to build the tool).

Three step loop that works for me 1. Generate tasks - Codex / Claude Opus → self-review.

2. Generate plan - same agent, “plan” mode → tweak if needed.

3. Implement - Claude Sonnet / Codex → review & merge.

For simple features I can even run this from my phone: ChatGPT app (Codex) → GitHub app → ChatGPT app → GitHub merge.

Repo: https://github.com/MrLesk/Backlog.md

Would love feedback and happy to answer questions!

13 comments

mrlesk

Reply
mitjam  6 days ago

Really love this.

Would love to see an actual end to end example video of you creating, planning, and implementing a task using your preferred models and apps.

  • mrlesk  6 days ago

    Will definitely do. I am also planning to run a benchmark with various models to see which one is more effective at building a full product starting from a PRD and using backlog for managing tasks

    • bazooka5798  6 days ago

      I'd love to see openRouter connectivity to try non Claude models for some of the planning parts of the cycle.

    • westurner  6 days ago

      Is there an established benchmark for building a full product?

      - SWE-bench leaderboard: https://github.com/FoundationAgents/MetaGPT :

      > Software Company as Multi-Agent System

      > MetaGPT takes a one line requirement as input and outputs user stories / competitive analysis / requirements / data structures / APIs / documents, etc. Internally, MetaGPT includes product managers / architects / project managers / engineers. It provides the entire process of a software company along with carefully orchestrated SOPs.

      - Mutation-Guided LLM-based Test Generation: https://github.com/codefuse-ai/Awesome-Code-LLM :

      > 8.2 Benchmarks: Integrated Benchmarks, Evaluation Metrics, Program Synthesis, Visually Grounded Program, Synthesis, Code Reasoning and QA, Text-to-SQL, Code Translation, Program Repair, Code Summarization, Defect/Vulnerability Detection, Code Retrieval, Type Inference, Commit Message Generation, Repo-Level Coding

      - underlines/awesome-ml/tools.md > Benchmarking: https://arxiv.org/abs/2402.00350

      2 replies →

beef_rendang  5 days ago

>ChatGPT app (Codex) → GitHub app → ChatGPT app → GitHub merge

I look forward to a future where we are reduced to rubberstamping fully-agentic-generated code on our glass slates for $0.01 eurodollars a PR.

thelittleone  6 days ago

I've had same experience. Taskmaster-ai was pretty good, but sometimes the agent ignored it as the project grew larger (can probably prevent that now using claude code hooks).

Trying this project today looks nice. I see you have sub-tasks. Any thoughts on a 'dependency' relation? I.e., don't do X if it is dependent on task A which is not complete.

FYI, there is a 404 in the AGENTS.md GEMINI.md etc pointing to a non existing README.md.

  • mrlesk  5 days ago

    Yep. Dependecies are supported via —dep parameter.

    Will check the 404 issues. Thanks for reporting it

unshavedyak  6 days ago

Would love more detail on your integration with claude. Are you telling claude to use backlog to plan X task? Feels like some MCP integration or something might make it feel more native?

Though i've not had much luck in getting Claude to natively use MCPs, so maybe that's off base heh.

  • mrlesk  6 days ago

    No mcp, just custom instructions.

    When you initialize backlog in a folder it asks you if you want to set up agent’s instructions like CLAUDE.md. It is important to say yes here so that Claude knows how to use Backlog.md.

    Afterwards you can just write something like: Claude please have a look at the @prd.md file and use ultrathink to create relevant tasks to implement it. Make sure you correctly identify dependencies between tasks and use sub tasks when necessary.

    Or you can just paste your feature request directly without using extra files.

    Feels a bit like magic

    • jwpapi  5 days ago

      How can I change from gemini to claude ?

      Also I’m not fully sure about your setup. In my fresh pov I would next set up agents that check my github repo for backlog tasks and do pull requests on those tasks. If I write a good description and ideally tests I can optimize the results of these.

      This creates the possibility of agents checking your backlog and prepare the work.

      I usually work with aider everyday and I’m quite fast in achieving task, the next limitation would be the latency and some back and forth. I have some dead time in between. I can definitely define tasks faster than 1-1 AI.

      Yeah if you could share a bit more how you do this with Claude we would all be thankful, also I havent seen anywhere to sponsor/tip you, would love to!