slogansand

5 ms·

1.
by slogansand 2mo ago
i have a claude subscription and i use https://github.com/asklokesh/loki-mode for most things, it's just working for me
2.
by slogansand 2mo ago
because it's all AI generated content. Unfortunately, no truth or human interest in reading it except what AI feels like with a gem of keyword manipulation of a million characters
3.
by slogansand 2mo ago
I have stopped autocompletions. Found easier to just prompt and get it done
4.
by slogansand 8mo ago
Install: bash # npm npm install -g loki-mode # Homebrew brew install asklokesh/tap/loki-mode # Docker docker pull asklokesh/loki-mode Usage: bash # CLI wrapper loki start ./prd.md loki start --provider codex ./prd.m
5.
by slogansand 8mo ago
What it does: Reads your PRD and breaks work into phases: Discovery, Architecture, Infrastructure, Development, QA, Deployment, and Growth Runs the RARV cycle (Reason, Act, Reflect, Verify) continuously Uses parallel agents for independent
6.
Loki Mode: Autonomous System for Claude, Codex and Gemini CLI(github.com)
1 points by slogansand 8mo ago|2 comments
7.
by slogansand 8mo ago
[2.35.0] - 2026-01-08 Added - Anthropic Agent Harness Patterns & Claude Agent SDK Sources: Effective Harnesses for Long-Running Agents - Anthropic Engineering Claude Agent SDK Overview - Anthropic Platform New Patterns: One Feature at a
8.
Show HN: Research-Backed Multi-Agent System for Autonomous Development(github.com)
3 points by slogansand 8mo ago|1 comments
9.
by slogansand 8mo ago
Last time someone raised concerns about web crawling for competitive research. Valid point. New version has configurable research modes. You can disable external crawling entirely and run fully offline if needed. Feedback heard.
10.
by slogansand 8mo ago
vs single-agent coding assistants: They tap out around 50% on SWE-Bench. No specialization. No parallel review. No self-healing. vs other multi-agent frameworks: Most focus on chat or simple task delegation. Loki Mode runs full SDLC - from
11.
by slogansand 8mo ago
For the skeptics (fair): FireLater repo has full git history. You can see the commits. No human intervention in the implementation phase. I reviewed outputs and approved deployments. But architecture decisions, code, tests, docs - all Loki
12.
by slogansand 8mo ago
On the swarm architecture for those curious: Engineering (8 types): frontend, backend, database, mobile, API, QA, perf, infra Operations (8 types): devops, SRE, security, monitoring, incident, release, cost, compliance Business (8 types): m
13.
by slogansand 8mo ago
Author here. Quick context on the benchmarks: We used RARV (Retrieve, Analyze, Reason, Validate) pattern with multi-agent collaboration. Each problem gets worked by specialized agents, reviewed by 3 parallel reviewers (code, business logic,
14.
Show HN: Loki Mode hit 99.67% SWE-Bench – MAF built a SaaS overnight(github.com)
2 points by slogansand 8mo ago|5 comments
15.
by slogansand 9mo ago
https://skillsmp.com/skills/asklokesh-claudeskill-loki-mode-... loki-mode in > Agent Skills Marketplace
16.
by slogansand 9mo ago
makes sense
17.
by slogansand 9mo ago
@boundring Appreciate the edit. And yeah, the crawling/consent question is thorny - the whole AI ecosystem is built on it. Thanks for checking it out. I'll revisit this and release a new version considering crawling.
18.
by slogansand 9mo ago
Fair question. What specifically worries you about it?
19.
by slogansand 9mo ago
I built this to solve my own problem - too many side projects dying because solo founders can't be 10 people at once. It's a Claude Code skill that orchestrates 37 specialized AI agents (engineering, ops, marketing, legal, etc.) t
20.
Show HN: Loki Mode – 37 AI agents that autonomously build your startup(github.com)
4 points by slogansand 9mo ago|8 comments