X @@akshay_pachaar · May 16, 2026 Full analysis by SuperBM

Akshay 🚀: the anatomy of ~/.hermes folder.

6/10 Solid

Explains the internal folder structure of the Hermes AI agent for customization.

Key Insights

  • Folder taxonomy separates identity, memory, skills, state, and automation clearly.
  • Agent self-creating skills during sessions is a novel but unproven claim.
  • Public GitHub repos as custom taps is a practical extensibility pattern.

Caveats & Flags

  • Akshay uses unverified, grandiose claims like '687 skills across 18 categories' without sourcing.
  • No evidence that a single folder controls 'everything' an agent knows—overstated.
  • Claims 'Soul.md occupies slot #1' with no technical citation or documentation link.

Valid Points

  • Folder structure description for skills, config, and state is internally consistent.
  • Memory consolidation via merging and truncation is a plausible design choice.
  • OAuth credentials stored in auth.json and API keys in env file follows common practice.

Counterpoints

  • No independent source confirms Hermes agent actually exists or uses this exact layout.
  • Claimed automation via cron and natural-language scheduling lacks evidence of reliability.
  • SQLite FTS5 indexing for tier 2 memory is plausible but unverified in this context.

Save this + 9 more analyses free

Your first save is this analysis

Sign in with Google →

Tag @superbmbot on Threads or @superbmHQ on X to analyze any post instantly

About this analysis

Is this claim legitimate?

SuperBM rates this content 6/10 (Solid). Explains the internal folder structure of the Hermes AI agent for customization.

What are the key issues with this content?

  • — Akshay uses unverified, grandiose claims like '687 skills across 18 categories' without sourcing.
  • — No evidence that a single folder controls 'everything' an agent knows—overstated.
  • — Claims 'Soul.md occupies slot #1' with no technical citation or documentation link.

What is actually useful in this post?

  • — Folder taxonomy separates identity, memory, skills, state, and automation clearly.
  • — Agent self-creating skills during sessions is a novel but unproven claim.
  • — Public GitHub repos as custom taps is a practical extensibility pattern.