@simpleapps-com/augur-skills 2026.3.9 → 2026.3.11

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@simpleapps-com/augur-skills",
-  "version": "2026.03.9",
+  "version": "2026.03.11",
   "description": "Install curated Claude Code skills",
   "license": "MIT",
   "type": "module",

package/skills/simpleapps/bash-simplicity/SKILL.md

@@ -30,6 +30,9 @@ Dedicated tools are faster, require no permission, and produce better output. MU
 
 Reserve Bash for commands that have no dedicated tool equivalent: build tools, test runners, git, package managers, and system commands.
 
+These commands are **denied** in project settings and will always be rejected — do not attempt them:
+`cd`, `cat`, `grep`, `rg`, `find`, `sed`, `awk`, `head`, `tail`, `sleep`, `kill`, `pkill`
+
 ## Check Before Prompting
 
 Before running a command that will trigger a permission prompt, check the wiki and project settings for approved commands. The wiki documents which commands are pre-approved and how to invoke them. Unnecessary permission prompts interrupt the user's flow.

package/skills/simpleapps/context-efficiency/SKILL.md

@@ -0,0 +1,70 @@
+---
+name: context-efficiency
+description: How to write token-efficient CLAUDE.md, rules, and skills. Use when creating or editing always-loaded content (CLAUDE.md, .claude/rules/) or authoring skills. Covers the context loading hierarchy, evergreen content principles, and platform limits.
+user-invocable: false
+---
+
+# Context Efficiency
+
+Always-loaded content (CLAUDE.md, rules without `paths`) is paid on every prompt — even when irrelevant. Every token spent here is a token taken from working context. Write always-loaded content as pointers, not content.
+
+## Context Loading Hierarchy
+
+| Layer | When loaded | What belongs here |
+|-------|------------|-------------------|
+| CLAUDE.md | Every prompt | Orient + link. Under 200 lines (official limit). |
+| Rules (no `paths`) | Every prompt | Short guardrails. One topic per file. Invoke a skill for detail. |
+| Rules (with `paths`) | On demand | Scoped guidance that only matters for specific file types. |
+| Skills (descriptions) | Every prompt | One-line descriptions. 2% of context window for ALL combined. |
+| Skills (full content) | When invoked | Full behavioral detail. Under 500 lines per skill. |
+| Wiki | When read | Complete project knowledge. Unlimited but we budget 20K tokens. |
+
+## Platform Limits
+
+- **CLAUDE.md**: under 200 lines. "Bloated files cause Claude to ignore your actual instructions."
+- **Skill descriptions**: 2% of context window for all skills combined (16K char fallback). Override with `SLASH_COMMAND_TOOL_CHAR_BUDGET`.
+- **SKILL.md**: under 500 lines. Move detail to supporting files.
+- **Auto memory**: first 200 lines of MEMORY.md loaded. Topic files on demand.
+- **Overhead**: ~31% of a 200K context window is consumed by system prompt, tools, and autocompact buffer before you type anything.
+
+## Evergreen Content
+
+Always-loaded content MUST be evergreen — true regardless of when it's read. MUST NOT contain:
+
+- **File counts or lists** — "12 skills" becomes wrong the next commit. Say "check `ls plugins/simpleapps/skills/`" instead.
+- **Version numbers** — "currently v2026.03.8" is stale immediately. Point to the VERSION file.
+- **Process data** — timestamps, recent activity, who did what. That's git history.
+- **Hardcoded paths** that change across machines or environments.
+- **Content that duplicates the code** — the code is the source of truth. Describe the pattern, link to the file.
+
+Instead: describe the **pattern**, give **1-2 examples**, and **link** to the source for the current state.
+
+## The Pointer Pattern
+
+Always-loaded files SHOULD be pointers that invoke on-demand content:
+
+```markdown
+# Git Safety (rule — 3 lines, always loaded)
+MUST NOT commit without user approval.
+Load Skill("git-safety") for full guardrails.
+```
+
+The rule costs ~40 tokens per prompt. The skill costs nothing until invoked, then loads ~800 tokens of detailed guidance. This is 20x more efficient than putting the full content in the rule.
+
+## When to Use Each Layer
+
+| Question | Answer |
+|----------|--------|
+| Does the agent need this on EVERY prompt? | CLAUDE.md or rule |
+| Is it a critical guardrail? | Rule (invokes skill for detail) |
+| Is it only relevant for certain files? | Rule with `paths` frontmatter |
+| Is it detailed behavioral guidance? | Skill |
+| Is it shared knowledge across projects? | Wiki |
+| Is it personal to one user? | Memory |
+
+## Checking Your Budget
+
+Run `/context` to see current token usage by category. Watch for:
+- Skills being excluded (descriptions exceeded 2% budget)
+- CLAUDE.md consuming more than ~5% of context
+- Many MCP servers inflating tool definitions

package/skills/simpleapps/wiki/SKILL.md

@@ -73,6 +73,21 @@ Pages MUST link to related sections on other pages using `[[Page-Name#section]]`
 
 Cross-linking also eliminates duplication. If two pages explain the same concept, one MUST become the source of truth and the other MUST link to it. Never duplicate content across pages — link instead. This keeps the wiki lean and ensures updates happen in one place.
 
+## Wiki Over Memory
+
+When asked to save, document, or record knowledge — use the wiki, MUST NOT use memory. The wiki is shared across all agents, all projects, and all computers. Memory is personal to one user on one machine — invisible to everyone else.
+
+| Knowledge type | Where it belongs |
+|---------------|-----------------|
+| Conventions, patterns, decisions | **Wiki** |
+| Learnings from a task or session | **Wiki** |
+| Architecture and process docs | **Wiki** |
+| How a problem was solved | **Wiki** |
+| Personal preferences (writing style, response length) | Memory |
+| User role and background | Memory |
+
+If in doubt, it belongs in the wiki. The cost of putting shared knowledge in memory is that it dies with the session — no other agent, project, or computer will ever see it.
+
 ## Keep It Lean
 
 - Document patterns and principles, not exhaustive lists