Pick the components you need.

You don't need all of PRISM. Every component is small, independent, and composable — a Bun-built dist/*.js bundle sharing one runtime. Start from your goal, install the minimal set, add more later. All live on the sovereign forge at git.sovereign-society.org/prism.

What do you actually need?

You want…	You need	You can skip
Steering files, a real git workflow, team dev skills	prism-harness	Everything else — the doctrine alone disciplines your agent.
Route across providers/models, rotate plans, safe keys	prism-route	loop, sober, proxy, chat — unless you want their layers too.
Cut token bills and audit context	prism-proxy	The rest — proxy compresses on its own.
Overnight memory; resumable agent	prism-loop	The rest — loop is self-contained durable memory.
Enforce your standards on every MR	prism-sober	The rest — sober is the review steering layer alone.
Just talk to your models — a lean, fast Hermes/OpenClaw-style chat, no agent in the loop	prism-chat	The harness and plugins — chat is a direct, efficient line to your models.

Want a lean, super-fast chat with your models and nothing else? prism-chat. Want your agent disciplined on top of that? Add prism-harness.

You probably don't need all six. The suite is a toolbox, not a requirement. Install the one that matches your goal above; every component works standalone.

The components, in detail

Each one answers: what do I get?

prism-harness — discipline

Any AI agent that visits your repo reads your doctrine before it writes a line: twelve rules, templates, six CLI subcommands (init, check, upgrade, openspec, steer, identity), plus the curl | sh meta-installer.

Get: one constitution for every agent you use. v1.0.0-beta.7

prism-route — routing + rotation

KDL profiles decide provider, account, and plan per request: 5h/weekly rotation, failover tiers, quota headroom scoring, file-based secrets from ~/.secrets/, Prometheus metrics at /metrics. Keys never enter env vars.

Get: multi-model routing; expensive plans rotate themselves. v1.0.0-beta.7

prism-loop — memory

File-based, lock-coordinated protocol for overnight agent loops: durable .prism/ memory, run boundaries, resumable across sessions and agents (Claude Code, Codex, OpenCode).

Get: overnight work that remembers until morning. v1.0.0-beta.7

prism-sober — reviews

Condenses your doctrine into byte-capped steering files Sober loads on every MR review (4,096 B per file, 12,000 B total). No LLM, no network — deterministic.

Get: your standards enforced on every merge request. v1.0.0-beta.7

prism-proxy — compression + audit

Loopback proxy ranks every tool by context-window consumption, then compresses: RTK for tool bloat, real WebP@87 vision with bold/italic for long prose, Caveman for the rest. Errors, stack traces, and results are always preserved.

Get: up to 70% smaller context bills, full audit trail. v1.0.0-beta.7

prism-chat — direct chat

Your lean, fast line to the models — a Hermes/OpenClaw-style chat with no agent in the loop. Terminal (ask/repl) plus a local installable PWA at 127.0.0.1:8789. Works standalone (point at a gateway), or rides prism-route serve for routing. Streaming, session usage, markdown export — no keys in the client.

Get: talk straight to your models, fast and efficient. v1.0.0-beta.7

Install — one line Browse the forge ↗

What the agent reads

The CLIs are small, but together they enforce twelve markdown rules and ship dozens of templates. On every session start, the visiting AI agent reads the doctrine at agents/rules/*.md first. That reading is what stops AI-slop:

Repo firewall — public/private boundary discipline.
Atomic forge cycle — every change is a complete, verifiable step.
Panopticum — one feature, one folder, one index.
TDD commit validation — no test, no commit.
SDD/BDD ordering — EARS for rigor, Gherkin for walkthrough.
Git worktree workflow — branch, worktree, rebase, push gates.
OpenSpec naming — v001- changesets, c001- capabilities, gapless sequences.
Crypto quality — cryptographic work without vectors is speculation.
SAP future-proofing — adapters over hard-coded SAP.
Customer disclosure — public/private content boundaries.
Agent execution — read context → identify slice → test → implement → verify → report.
Goals — primary goal + current priority + decision rule.

Path conventions — non-overlapping

your-project/agents/{rules,templates,scripts,...}      ← doctrine (per repo)
your-project/.prism/{blueprint,ledger,heartbeat,...}    ← loop memory (per repo)
~/.prism/identity/*.md                                  ← agent identity (per user)
~/.sober/steering/*.md                                  ← Sober-Raccoon (per repo)
~/.config/containers/systemd/prism-*.container         ← container (per host)

Each tool owns its own surface. No path overlap. Each tool has a sovereign index.

Model + services first

The installer prompts component selection in priority order: model + services (prism-route) first, plugins after. Without prism-route the identity canon has nothing to talk through. Without prism-sober the doctrine is unverifiable. Without prism-loop the agent forgets by morning. Each adds a layer. And prism-chat hands the same gateway to the human directly — a terminal and a PWA, no agent in the loop.