Hi — submitting Perseus for inclusion. Happy to convert this to a PR if you prefer; opening as an issue because the README file is large and a clean PR diff is easier for me to write if you can confirm the placement first.
Suggested placement
Section: ## 🛠️ Tools & Utilities (at the bottom — secondarily fits 🏗️ Infrastructure & Proxies)
Suggested entry
- [**perseus**](https://github.com/tcconnally/perseus) (new) - Live context engine. Resolves `@query`, `@services`, `@waypoint`, and 19 more directives into the markdown file Claude Code reads at session start. Compile-before-context as an alternative to runtime MCP calls. `perseus install --target claude-code` drops SessionStart hooks. MIT, ~600 tests.
What it is
Perseus is a pre-processor that resolves environment-state directives inside .perseus/context.md and writes the result into whichever file the assistant reads at session start (CLAUDE.md, AGENTS.md, .cursorrules, .hermes.md). The assistant never sees directive syntax — it sees a document of verified facts.
Why this list specifically: Perseus is built around the Claude Code workflow (hook installer for .claude/settings.json, dedicated claude-md output format) but is multi-assistant by design.
Quick facts
Benchmark hooks (for the curious)
- 50,000 directives in 1.36s warm with
@cache — 450× cold→warm gap
- 120-agent coordination via atomic O_CREAT|O_EXCL checkpoint locking
- Enterprise simulation: 500 devs × 5-day workweek = 16,250 renders in 961s, vs ~83 hours equivalent via runtime LLM tool calls
Happy to send a PR once you confirm the section. Thanks for maintaining this list — it's how I discovered ~5 of the tools I now use daily.
Hi — submitting Perseus for inclusion. Happy to convert this to a PR if you prefer; opening as an issue because the README file is large and a clean PR diff is easier for me to write if you can confirm the placement first.
Suggested placement
Section:
## 🛠️ Tools & Utilities(at the bottom — secondarily fits🏗️ Infrastructure & Proxies)Suggested entry
What it is
Perseus is a pre-processor that resolves environment-state directives inside
.perseus/context.mdand writes the result into whichever file the assistant reads at session start (CLAUDE.md,AGENTS.md,.cursorrules,.hermes.md). The assistant never sees directive syntax — it sees a document of verified facts.Why this list specifically: Perseus is built around the Claude Code workflow (hook installer for
.claude/settings.json, dedicatedclaude-mdoutput format) but is multi-assistant by design.Quick facts
pip install perseus-ctxBenchmark hooks (for the curious)
@cache— 450× cold→warm gapHappy to send a PR once you confirm the section. Thanks for maintaining this list — it's how I discovered ~5 of the tools I now use daily.