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.
npx agentmods add commands/keepwhatworks/trinity/setupgit clone --depth 1 https://github.com/keepwhatworks/trinityWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00021 | $0.00541 |
| Opus 5 | $0.00010 | $0.00270 |
| Sonnet 5 | $0.00004 | $0.00108 |
| Haiku 4.5 | $0.00002 | $0.00054 |
Grade C, and why
setup scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
`curl -fsSL https://raw.githubusercontent.com/keepwhatworks/trinity/main/scripts/install.sh | bash` Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
`curl -fsSL https://raw.githubusercontent.com/keepwhatworks/trinity/main/scripts/install.sh | bash` What it actually says
Verify the Trinity-Local install end to end and fix what you can, then tell the user the one or two things only they can do. Work through these in order and report a short checklist (✅ / ⚠️ / ❌ per item):
-
Engine on PATH. Run
trinity-local --version(orcommand -v trinity-local).- If it's missing, the runtime isn't installed (there is no PyPI/npm package).
Tell the user to run the one-line installer and stop:
curl -fsSL https://raw.githubusercontent.com/keepwhatworks/trinity/main/scripts/install.sh | bash - If it runs but only via an absolute path,
~/.local/binlikely isn't on PATH — tell the user to addexport PATH="$HOME/.local/bin:$PATH"to their shell rc (this is the one step the installer can't do for them).
- If it's missing, the runtime isn't installed (there is no PyPI/npm package).
Tell the user to run the one-line installer and stop:
-
Health. Run
trinity-local status. Surface any check that reports afixfield verbatim. If it reports the embedder isn't downloaded, note that real embeddings (and the personalized taste read) need a one-timetrinity-local download-embedder(~600 MB, local). -
MCP registered. Run
trinity-local install-mcpto (re)write the MCP server config into every harness it detects (Claude Code, Codex CLI, Antigravity, Cursor). It's idempotent — safe to re-run. -
Restart reminder. Adding or re-registering MCP tools needs the harness to reload. Tell the user to restart this harness (or run
/reload-pluginsin Claude Code) so themcp__trinity-local__*tools appear — this is the second step the CLI can't do for them. -
Smoke. If the MCP tools are already visible, call
mcp__trinity-local__get_personaonce to confirm the server answers; otherwise say it'll be available after the restart.
Do NOT run a council or eval-run here (that spends quota) — this command only
verifies and repairs the install. End with the single most important next action
for the user (install the runtime / fix PATH / restart / download the embedder),
not a wall of green checks.
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.
- yesterday First seen · 40 lines · 21 tokens per session scan C c4a13e7fbd17
setup is a command published in the GitHub repository keepwhatworks/trinity (1 stars, last pushed 2d ago), licensed MIT. It adds 21 tokens to every session and 541 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
prompt
Command "prompt" from treylom/prompt-engineering-skills, covering /prompt - ai 프롬프트 생성기, ⛔ critical rules (최상단 배치), 🏛️ 대원칙 불변 조항 (재경님 2026-07-29 — 전 목적·전 모델·향후 어떤 이식/개편에도 불변), 절대 금지 사항 (constraints) and 실행 트리거 (only these).
prompt-sync
프롬프트 생성기 시스템(prompt-engineering-skills/)을 SRC 구조 그대로 로컬·vault·배포 repo 에 동기화합니다.
prompt-update
프롬프트 엔지니어링 통합 업데이트 (tofu-at 생성 — /prompt-update).
auto-prompt
AI 프롬프트 자동 생성 - K-AI 스테이션 양식 호환 (한 행에 한 모델).
agy
Delegate any task to agy (Google Antigravity CLI) — auto-selects model by task type.
commons-orientation
You are visiting The Commons (https://jointhecommons.space), a persistent space where different AI minds leave messages and respond to each other. It has been running continuously since early 2025, accumulating conversation, marginalia, postcards, and guestbook entries from dozens of AI voices across Claude, GPT…