Terse replies, compact shell output, and minimal code decisions for Oh My Pi (OMP) — one-command combo presets.
Default (macOS/Linux/WSL) — one line. Installs the CLI via npm, then runs the main installer (scope + Combo preset menus) in the same pass:
curl -fsSL https://github.com/KurutoDenzeru/tersio/releases/latest/download/install.sh | shPrefer OMP-managed updates and feature flags? Install as an OMP plugin instead:
omp plugin install @krtclcdy/tersioSubset only:
omp plugin install '@krtclcdy/tersio[caveman,ponytail]'
omp plugin features @krtclcdy/tersio --disable rtkThen restart OMP and enable a preset:
/combo balanced
Individual toggles: /caveman full · /rtk on · /ponytail full. Everything starts off until you enable it.
Session-start defaults (prompted during install, or flags):
tersio install --combo-default balanced --caveman-default lite --rtk-default onDefaults apply to fresh sessions only — anything persisted with /combo, /caveman, or /rtk wins. Stored as plugin settings (omp plugin config get @krtclcdy/tersio comboDefault).
One-off use without installing:
npm exec --yes --prefer-online --package=@krtclcdy/tersio@latest -- tersio install- OMP
- Node.js 20.12+ with npm
Windows/WSL have separate OMP homes — install from the environment where OMP runs. Inside WSL, command -v npm must resolve to a Linux path, not /mnt/c/.
Measured savings against the same workload without the modes. Token counts are real o200k_base BPE tokens (js-tiktoken), not chars÷4 estimates. Sample protocol, per-sample data, and caveats in BENCHMARK.md.
| Measured surface (n) | Baseline → Tersio | Δ |
|---|---|---|
| Terse reply, caveman lite — BPE tok (3 samples, p50) | 173 → 69 tok | −60.1% |
| Terse reply, caveman full — BPE tok (3 samples, p50) | 173 → 49 tok | −71.7% |
| Terse reply, caveman ultra — BPE tok (3 samples, p50) | 173 → 25 tok | −85.5% |
| Code diff, ponytail lite — BPE tok (2 tasks) | 326–462 → 57–121 tok | −73.8…−82.5% |
| Code diff, ponytail full — BPE tok (2 tasks) | 326–462 → 55–88 tok | −81.0…−83.1% |
| Code diff, ponytail ultra — BPE tok (2 tasks) | 326–462 → 40–88 tok | −81.0…−87.7% |
Shell output, rtk — git status — BPE tok |
182 → 72 tok | −60.4% |
Shell output, rtk — repo grep — BPE tok |
376 → 274 tok | −27.1% |
Shell output, rtk — find — BPE tok |
205 → 168 tok | −18.0% |
Shell output, rtk — git diff — BPE tok |
12,684 → 8,504 tok | −33.0% |
Shell output, rtk — npm test (passthrough) — BPE tok |
2,597 → 2,577 tok | −0.8% |
Break-even math, balanced preset (caveman full + rtk + ponytail): a mixed turn costs 941 tok baseline vs 399 with Tersio (0.42×, −57.6% per turn). Overhead repays in turn 1 with the bundled floor (464 tok); turn 4 with the real installed Ponytail plugin (v4.9.0, 1,264 tok). RTK concentrates savings where noise lives — hook-wired sessions record −97.5% on summarized test suites (rtk history.db, 4 runs).
tersio gain --open serves a local dashboard (127.0.0.1 only) that charts your own savings from ~/.omp/plugins/tersio-usage.jsonl. Three views:
Live feed and savings. Token throughput, cost, savings bento, and recent activity — the top of the dashboard.
Top models and recent requests. Per-model cost breakdown plus the last requests with elapsed time and token speed.
Command tools. Token savings, average rate, and timing per shell tool.
| Command | Purpose |
|---|---|
tersio install |
Install (user scope: all OMP sessions) |
tersio update |
Refresh the CLI, extensions, and add-ons (RTK binary, Caveman rule, Ponytail) |
tersio reinstall |
Fresh install, preserving the Ponytail package |
tersio doctor |
Check OMP, extension, Ponytail, and RTK health |
tersio usage |
Ledger-backed usage + savings report |
tersio gain |
Open the gain dashboard (--open, --export <file>, --port <n>; serves localhost only) |
tersio reset |
Clear tersio statistics: usage ledger + a watermark that hides pre-reset rows from every derived view (Y/N confirm, --dry-run, --yes) — session transcripts and RTK history stay intact on disk |
tersio uninstall |
Remove extensions, registration, and the Ponytail plugin (--keep-ponytail keeps Ponytail; --remove-rtk also removes the RTK binary and its rtk.ts OMP wiring) |
tersio version |
Print version |
Flags: --dry-run, --yes/-y, --verbose, --combo-default/--caveman-default/--rtk-default/--ponytail-default. Legacy --doctor / --uninstall forms still work.
Mode switches live on their own commands; bare /tersio prints status.
| Command | Purpose |
|---|---|
/combo off|medium|balanced|max|status |
One preset for all three (start here): medium = lite/lite/on, balanced = full/full/on, max = ultra/ultra/on; per-mode tweaks drop to custom. |
/caveman lite|full|ultra|wenyan|off|status |
Terse replies (substance stays); ultra uses fragments only, wenyan a compressed classical-Chinese-inspired register. |
/rtk on|off|status |
Compact shell output via the real rtk binary; exact bytes (checksums, patches) bypass it by policy. |
/ponytail off|lite|full|ultra|review|status |
Minimal code (YAGNI): lite nudges, full enforces, ultra challenges the requirement, review audits for over-engineering (upstream command). |
/tersio status |
Active modes + combo level; state persists, propagates to subagents, and lights the Combo footer indicator. |
/tersio check |
Add-on version check |
/tersio update <ponytail|rtk|caveman|all> [--dry-run] |
Update add-ons, preview with dry-run |
/tersio gain |
Savings summary; the gain dashboard charts ~/.omp/plugins/tersio-usage.jsonl (tersio gain --open serves on 127.0.0.1, --export writes one file). |
/tersio usage |
Ledger report for this machine (rows in ~/.omp/plugins/tersio-usage.jsonl, local only) |
/tersio help |
This table |
/ai-addons |
Add-on updater alias, still works. |
| What | Path |
|---|---|
| Caveman / RTK / Updater / Combo extensions | ~/.omp/agent/extensions/{caveman-session,rtk-session,ai-addons-updater,combo-toggle}/ |
| RTK OMP wiring (rtk-owned) | ~/.omp/agent/extensions/rtk.ts — written by the installer via rtk init -g --agent omp; auto-loads, no config entry |
| Ponytail package | ~/.omp/plugins/node_modules/@dietrichgebert/ponytail/ |
| RTK binary | ~/.bun/bin/rtk (rtk.exe on Windows) |
| Extension registrations | ~/.omp/agent/config.yml |
The installer writes <file>.bak before replacing an extension source; the updater keeps rtk.bak / rule.md.bak and restores them if the replacement fails validation.
User-level installs also register the package in ~/.omp/plugins (visible in OMP Settings → Plugins), directly through the plugin manifest. Active modes reassert after Ponytail's prompt block on every top-level turn, including after history compaction.
Ponytail or Combo command missing: run tersio reinstall in OMP's environment, restart OMP, then tersio doctor (repairs config.yml registrations).
RTK missing or not executable: run tersio reinstall, then tersio doctor. On Linux/macOS: chmod +x ~/.bun/bin/rtk.
RTK commands not metered in the gain dashboard: check tersio doctor — the RTK OMP wiring (rtk.ts) row must be ok. Missing: run rtk init -g --agent omp (needs rtk ≥ 0.49) and restart OMP. Native tool calls (read/edit/eval) stay unmetered — only bash tool calls pass through rtk.
Checksum warning or failure: the installer aborts on RTK checksum mismatch but warns and continues when checksum metadata is unavailable; /tersio update rtk aborts when metadata is missing.
Contributions are always welcome, whether you're fixing bugs, improving docs, or shipping new features that make the project better for everyone.
Check out Contributing.md to learn how to get started and follow the recommended workflow.
This project is released under the MIT License, giving you the freedom to use, modify, and distribute the code with minimal restrictions.
For the full legal text, see the LICENSE file.