Love the concept and agree it can become a thing. On the schema, not sure about skillflows, tools, knowledge, memory - those aren't much standardized today, but agree they help as additional primitives, and standardizing would help.
I built Agent Package Manager at Microsoft - wondering if it may supercharge GitAgents with dependency modules so that they can become composable, same as for classic software. Many common core ideas on the paradigm, curious on your take https://github.com/microsoft/apm
Similar idea (rooted on "agents as markdown") but on the outer loop is taking shape at GitHub with GitHub Agentic Workflows.
The trace analysis is the most interesting part of this paper.
Agents dutifully follow AGENTS.md instructions — they run more tests, grep more files, do more checks — but that thoroughness costs 19% more inference without meaningfully improving outcomes. That's not an argument against context files, it's evidence we're writing them wrong.
We're authoring them like READMEs for humans ("here's the architecture, here's how we structure things") when agents actually need narrow, non-inferable directives — the custom build flag, the weird test harness, the constraint they'd waste 30 tool calls discovering.
The HN commenter who noted the real value is "forcing you to articulate things previously just in your head" is actually pointing at the same conclusion from the other direction: the thinking matters, but the artifact needs a different shape than what we've converged on.
Scope instructions to file patterns (services/* → "use ServiceFoo, no raw queries"). The agent only loads rules relevant to the code it's touching, and a small scoped file is easier to notice is wrong than line 47 of a monolithic doc.
Add a post-edit agent hook that asks the agent "does this change mean the scoped instructions need updating?" — turns maintenance from a chore someone forgets into part of the edit loop. Haven't battle-tested this at scale yet but the mechanism is there in Claude Code and Copilot.
For bootstrapping the initial instructions, check out microsoft/agentrc — it can generate and evaluate scoped instruction files so you're not writing them from scratch.
I built Agent Package Manager at Microsoft - wondering if it may supercharge GitAgents with dependency modules so that they can become composable, same as for classic software. Many common core ideas on the paradigm, curious on your take https://github.com/microsoft/apm
Similar idea (rooted on "agents as markdown") but on the outer loop is taking shape at GitHub with GitHub Agentic Workflows.