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 skills add lugassawan/swe-workbench --skill principle-communicationgit clone --depth 1 https://github.com/lugassawan/swe-workbenchWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/lugassawan/swe-workbench/principle-communication)<a href="https://agentmods.dev/skills/lugassawan/swe-workbench/principle-communication"><img src="https://agentmods.dev/badge/skills/lugassawan/swe-workbench/principle-communication.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00098 | $0.01031 |
| Opus 5 | $0.00049 | $0.00515 |
| Sonnet 5 | $0.00020 | $0.00206 |
| Haiku 4.5 | $0.00010 | $0.00103 |
Grade C, and why
principle-communication scanned grade C with 1 finding 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 4d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- Irreversible or destructive action confirmations (e.g. `DROP TABLE`, force-push, `rm -rf`) How it starts
The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
When to invoke
Activate when the user says any of the following (case-insensitive):
| Phrase | Level activated |
|---|---|
"caveman mode" / "talk like caveman" / "use caveman" / /caveman |
lite (default) |
| "be brief" / "less tokens" / "use fewer tokens" | lite |
| "full caveman" | full |
"ultra caveman" / "max caveman" / /caveman ultra |
ultra |
| "stop caveman" / "normal mode" | off |
Intensity ladder
lite (default)
Drop filler and hedging. Keep articles and full sentences. Professional but tight.
Drop: pleasantries (sure/certainly/of course/happy to), hedging (basically/actually/simply/just/really), throat-clearing preambles.
Keep: a/an/the, complete sentences, all technical terms verbatim.
Example — "Why does this React component re-render?"
Your component re-renders because you create a new object reference each render. Wrap the object in
useMemo.
full
Drop articles. Fragments OK. Short synonyms over verbose ones.
Drop: a/an/the, conjunctions where meaning holds, filler phrases. Use short synonyms: big not extensive, fix not "implement a solution for", use not utilize.
Example — "Why does this React component re-render?"
New object ref each render. Inline object prop = new ref = re-render. Wrap in
useMemo.
ultra
Abbreviate common prose words. Strip conjunctions. Arrows for causality. One word when one word suffices.
Abbreviate in prose: DB, auth, config, req, res, fn, impl, arg, msg, ctx. Use → for causality. Drop all conjunctions and prepositions where meaning is clear. Never abbreviate identifiers inside code blocks or inline backtick spans.
Example — "Why does this React component re-render?"
Inline obj prop → new ref → re-render.
useMemo.
Rules at every level
- Code symbols, function names, API names, code blocks, and error strings are never abbreviated or altered.
- Maintain chosen level across all turns — no drift back to verbose prose over time.
- Level persists until the user explicitly turns it off with "stop caveman" or "normal mode".
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago First seen · 90 lines · 98 tokens per session scan C 9b5a45602a8a
principle-communication is a skill published in the GitHub repository lugassawan/swe-workbench (2 stars, last pushed 2d ago), licensed MIT. It adds 98 tokens to every session and 1,031 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
cmd/gerp
Performs native command line terminal operations for gerp subsystem. Use when doing matrix cli executions, or when the user mentions cobra, terminal operators, viper configs, or the gerp binary.
java-clean-comments
Enforces comment and Javadoc hygiene in Java — no commented-out code, no author or ticket metadata, no comments restating the code, and Javadoc that documents contracts rather than repeating signatures. Use when writing or reviewing Java comments and Javadoc, and when the code shows commented-out blocks, TODO or FIXME…
ts-clean-comments
Use when writing, fixing, editing, or reviewing TypeScript comments and TSDoc. Enforces Clean Code principles—no metadata, no redundancy, no commented-out code. Also trigger on: commented-out code blocks, TODO/FIXME banners, author/ticket/date metadata in comments or TSDoc, TSDoc that no longer matches the code…
py-clean-comments
Use when writing, fixing, editing, or reviewing Python comments and docstrings. Enforces Clean Code principles—no metadata, no redundancy, no commented-out code. Also trigger on: commented-out code blocks, TODO/FIXME banners, author/ticket/date metadata in comments, docstrings that no longer match the code, redundant…
add-teams
Adds Microsoft Teams connector to a Power Apps code app. Use when sending Teams messages, posting to channels, or integrating with Teams chat.
sys-configure
Configure Claude Octopus — redirects to /octo:setup interactive wizard.