praxis-cli CLAUDE.md

Repository instructions for praxis-cli, a small Go command-line program that connects local AI hosts to Praxis cloud services. They explain its single-profile default and how profiles and server-side credentials should work.

In plain words
What is it for?
Use them when changing profile handling, command-line flags, skill fetching, or the HTTPS connection between local AI hosts and Praxis cloud deployments.
Why use it?
They help maintain the project’s intended simple default flow while keeping advanced multi-profile behavior from affecting ordinary users.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/facets-cloud/praxis-cli/claude-md
Clone the repo
git clone --depth 1 https://github.com/Facets-cloud/praxis-cli
Per session 4,765 This file is loaded in full into every session.
When invoked 4,765 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.04765 $0.04765
Opus 5 $0.02382 $0.02382
Sonnet 5 $0.00953 $0.00953
Haiku 4.5 $0.00477 $0.00477

Measured 2d ago against content hash 813d5924a434, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

praxis-cli CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 357 lines

How it starts

The opening of the file, as written. The whole thing — 357 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — praxis-cli

Guidance for Claude Code (and other AI assistants) working in this repo. Project-specific overrides for the global ~/.claude/CLAUDE.md.

Project overview

Single-binary Go CLI (praxis) that exposes Praxis cloud capabilities to any local AI host (Claude Code, Cursor, Gemini CLI). The CLI is a thin HTTPS client to a Praxis cloud deployment — it does not run an agent loop locally. Skills are sourced (fetched + nomenclature-translated) into the user's AI host; MCP tools execute server-side under org-managed credentials. See README.md for the user-facing story.

Design principle — single-profile users first

The typical praxis user has ONE profile (a customer on one control plane); everything must resolve silently to "default" for them. Multi-profile users (Facets engineers, support) are power users: their flows must work, but they are strictly secondary and must never regress the single-profile flow. Concretely:

  • New flags and subcommands are opt-in; the default flow never grows required steps, prompts, or warnings a single-profile user would see.
  • Power-user affordances (profile pins, raptor cross-checks, per-directory profiles) live behind flags or status fields that stay inert when only one profile exists.
  • When a trade-off pits multi-profile ergonomics against single-profile cleanliness, single-profile wins.

Match this repo, and keep it small

Before adding anything, open 2-3 existing files that do the same job and copy their shape. A diff that reads like a different person wrote it is a review cost even when the logic is right.

  • Smallest diff that works. Fewest files, fewest new packages. Extending an existing file beats a new one; a new one beats a new package.
  • Comments: match the neighbours, which are terse. One line in a function body. Package/function docs 3-8 lines, one long block per file at most. Never explain what the code says, never defend a design — that goes in the PR body.
  • No layer for data we already have. Check what the server already sends and what the CLI already parses before writing a parser.
  • No speculative entries. A filter list, config key or flag with nothing using it today is dead state; leave it out.

Read the full file on GitHub · 357 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 357 lines · 4,765 tokens per session scan A 813d5924a434

Subscribe to this mod's changes

praxis-cli CLAUDE.md is an instructions file published in the GitHub repository Facets-cloud/praxis-cli (2 stars, last pushed 2d ago), licensed MIT. It adds 4,765 tokens to every session, about $0.0238 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.