@Everlier

Agent harness engineer @viktor_com. Built Harbor (3k★ local LLM stack). I test what actually works with coding agents and post the data.

av.codes
Joined April 2010
  • Tweets12.6K
  • Following467
  • Followers1.9K
1.6KMedia
You no longer carve software through the pain an sweat, you pour it now.

Software is a liquid now

Article · @Everlier
7
9
2
63
20,921
6/ TraceQuest is Open Source, runs on your machine and sends no telemetry. github.com/viktor-com/traceq…
9
3/ TraceQuest shows where runs fail or lose time without reading raw logs, failures are grouped by cause, error rate per agent, project, retries and loops, active time and estimated spend.
1
4
4/ With your sessions indexed in TraceQuest, your agent can write you standup notes, each tied to specific session.
1
2
5/ If you also want a UI, TraceQuest includes a local web app and can run as a daemon. You can watch (and analyse) while your agents are still working.
1
8
2/ TraceQuest indexes sessions from Claude Code, Codex, Cursor, OpenCode, Grok, Droid. You can configure TraceQuest to collect session from your fleet over SSH, as well as Cursor Cloud.
1
15
Incidentally, this is also best Grok 4.7 computer use to date, just saying ;)
SpaceX Super Intelligence
@SpaceXSIOct 21
Use Grok 4.7 in Viktor
1
116
We're prompting agents, but we are not the same
88
A skill that turns any agent/harness into ultracode from Claude. The agent will: - Split your work into steps - Arrange them into a graph (could be cyclic) - Orchestrate shared state on the filesystem - Continue running the graph until it's fully done. The largest run I ever had was ~1400 nodes. github.com/av/skills/tree/ma…
1
1
127
It started with one simple chat window. Then you ran one session, then a few at once. Now you talk to dozens of agents all day. How do you scale that without losing track of them?

You are a toll booth clerk for your coding agents

Article · @Everlier
2
2
11
425
That was good to see in the morning
1
96
Is there such thing as opting out of A/B tests with an LLM provider?
4
136
- 98% of my agent sessions are run by other agents, not me - about a third of my sessions are agents reviewing other agents work - a typical session makes 123 tool calls Personal trace intelligence
1
129
Sea glass meant to be the actual UI there. You interact with stones, and your agents do things. As simple as three seashells. av.codes/stones
SLK15@SLKVinylOct 9
I think Google could find success in making sea glass the next UI material for Material Design 4
2
232
Claude: I wrote some tests to simulate an environment where this bug never existed Me: oh
6
186
My agentic coding went from a chatgpt window to an agentic fleet It'll be an armada soon
1
1
6
239
music is "Digital Lemonade" by Kevin MacLeod
1
30
I don't recognize Opus 5.5 behavior today.
2
4
835
A colleague asked me today "what is your typing speed?", I replied "In tokens per second?" Turns out: Claude Fable 5.1: 58 tokens/s Me: 1.64 tokens/s docs.viktor.com/human-vs-llm
4
8
362
Use your agents to personalize the content for yourself. Make it enjoyable for you to consume. You'll be delighted. Some random examples I did in the past few days:
1
1
11
687
All software will be Open Source soon.
Amjad Masad
@amasadOct 7
What’s happening in the AI-powered reverse engineering and decompilation is absolutely insane. Pretty soon all software will be de facto open-source. AI is coming for everything and everyone.
2
1
1
11
612
Load more