omnilearn-workflow 1.0.2 → 1.0.4

package/commands/omnilearn-init.md

@@ -37,9 +37,20 @@ This global location is always accessible regardless of which project directory
 
 ```
 {learningDirectory}/
-└── .omnilearn/
-    ├── config.json              ← Mirror of the global config (for reference)
-    └── UserPreferences.md       ← Global user preferences (auto-populated)
+├── .omnilearn/                  ← ONLY configuration and global preferences
+│   ├── config.json              ← Mirror of the global config (for reference)
+│   └── UserPreferences.md       ← Global user preferences (auto-populated)
+│
+├── <skill>/                     ← Skills live at the ROOT of learning directory
+│   ├── roadmap.md
+│   ├── SkillPreferences.md
+│   ├── SkillConventions.md
+│   ├── progress-index.md
+│   ├── runs/
+│   └── topics/
+│
+└── <another-skill>/             ← Each skill is its own top-level folder
+    └── ...
 ```
 
 ## Phase 0: CHECK EXISTING CONFIG
@@ -117,7 +128,8 @@ sed -i "s|DATE_PLACEHOLDER|$(date +%Y-%m-%d)|g" "$OMNILEARN_CONFIG"
 ### 2.2 Create Base Directory Structure
 
 ```bash
-# Create .omnilearn in the learning directory
+# Create .omnilearn in the learning directory (config only)
+# Skills will be created as top-level directories at $LEARNING_DIR
 OMNILEARN_DIR="$LEARNING_DIR/.omnilearn"
 mkdir -p "$OMNILEARN_DIR"
 

package/commands/omnilearn-refine.md

@@ -25,7 +25,7 @@ Both scenarios follow the same core flow: research → analyze → produce struc
 ## Directory Structure
 
 ```
-.omnilearn/<skill>/
+<skill>/                                          ← Skill at learning directory root
 ├── roadmap.md
 ├── SkillPreferences.md
 ├── SkillConventions.md                  ← 🔑 Setup conventions read for consistent diagnostic tasks
@@ -56,9 +56,9 @@ CURRENT_YEAR=$(date +%Y)
 
 All subagents that use MCP tools MUST follow these exact calling conventions:
 
-### `context7_resolve-library-id` + `context7_query_docs`
+### `context7_resolve-library-id` + `context7_query-docs`
 1. **ALWAYS call `context7_resolve-library-id` FIRST** with the library name to get the correct library ID.
-2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query_docs`.
+2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query-docs`.
 3. Do NOT guess or hardcode library IDs.
 4. Max 3 calls per question.
 
@@ -79,7 +79,7 @@ All subagents that use MCP tools MUST follow these exact calling conventions:
 | Tool | When | Why |
 |------|------|-----|
 | `google_search` / `websearch_web_search_exa` | Research | Topic research, answer questions, validate changes |
-| `context7_resolve-library-id` + `context7_query_docs` | Tech topics | Official documentation for precise answers — MUST call resolve first |
+| `context7_resolve-library-id` + `context7_query-docs` | Tech topics | Official documentation for precise answers — MUST call resolve first |
 | `task(category="unspecified-high", background)` | Heavy work | Research, analysis, topic updates |
 | `read`, `write`, `edit`, `bash`, `grep`, `glob` | All phases | File operations |
 
@@ -121,7 +121,7 @@ fi
 
 LEARNING_DIR=$(grep -o '"learningDirectory"[[:space:]]*:[[:space:]]*"[^"]*"' "$OMNILEARN_CONFIG" | sed 's/"learningDirectory"[[:space:]]*:[[:space:]]*"//' | sed 's/"$//')
 OMNILEARN_DIR="$LEARNING_DIR/.omnilearn"
-SKILL_DIR="$OMNILEARN_DIR/<skill>"
+SKILL_DIR="$LEARNING_DIR/<skill>"
 TOPIC_DIR="$SKILL_DIR/topics/<topic>"
 TOPIC_ROADMAP="$TOPIC_DIR/topic-roadmap.md"
 CURRENT_RUN="$SKILL_DIR/runs/$(date +%s)-refine-<topic>"
@@ -162,7 +162,7 @@ task(category="unspecified-high", run_in_background=false, prompt="
 1. TASK: Create a focused micro-exercise that tests the user's understanding of '{concept}' in '{skill}'. The user is confused about: '{user_request}'.
 2. EXPECTED OUTCOME: A self-contained micro-exercise (5-15 min) that isolates the specific concept and lets the user figure it out by coding/building.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query-docs, read, write
 
 4. MUST DO:
    - Read the current topic-roadmap.md to understand what's been covered: {TOPIC_ROADMAP}
@@ -275,7 +275,7 @@ task(category="unspecified-high", run_in_background=true, prompt="
 1. TASK: Research online to find better ways to structure and teach '{topic}' in '{skill}'.
 2. EXPECTED OUTCOME: A research document with web-sourced findings about best practices for teaching/learning this topic.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query-docs, read, write
 
 4. MUST DO:
    - Analyze the user's refinement request: '{user_request}'
@@ -410,7 +410,7 @@ Write to `$CURRENT_RUN/agent-log.md`:
 
 ```bash
 if git rev-parse --git-dir > /dev/null 2>&1; then
-  git add "$OMNILEARN_DIR/"
+  git add "$LEARNING_DIR/"
   git commit -m "omnilearn: refine topic '{topic}' in {skill}
 
 - {summary of what was done}
@@ -442,7 +442,7 @@ If no git repo: ask if user wants to initialize one (same pattern).
 | Topic not found | Suggest available topics from the skill roadmap |
 | User's question is too vague | Ask clarifying questions before researching |
 | Research yields poor results | Try alternative search queries, broaden scope |
-| Refinement accidentally modifies assignments | Revert: `git checkout .omnilearn/{skill}/topics/{topic}/assignments/` |
+| Refinement accidentally modifies assignments | Revert: `git checkout $SKILL_DIR/topics/$TOPIC/assignments/` |
 | User wants to undo refinement | Archive current, restore from git or previous file in runs/ |
 | User has multiple questions | Answer the most foundational one first, then the dependent ones |
 

package/commands/omnilearn-roadmap-edit.md

@@ -25,9 +25,9 @@ CURRENT_YEAR=$(date +%Y)
 
 All subagents that use MCP tools MUST follow these exact calling conventions:
 
-### `context7_resolve-library-id` + `context7_query_docs`
+### `context7_resolve-library-id` + `context7_query-docs`
 1. **ALWAYS call `context7_resolve-library-id` FIRST** with the library name to get the correct library ID.
-2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query_docs`.
+2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query-docs`.
 3. Do NOT guess or hardcode library IDs.
 4. Max 3 calls per question.
 
@@ -59,14 +59,15 @@ This command performs the same rigorous research-backed process as `/omnilearn-r
 ## Directory Structure Reference
 
 ```
-.omnilearn/
-├── UserPreferences.md
-└── <skill-name>/
+.omnilearn/                                        ← Config only
+└── UserPreferences.md
+
+<skill-name>/                                      ← Skill at learning directory root
     ├── SkillPreferences.md
     ├── SkillConventions.md                  ← Preserved during edits
     ├── roadmap.md                          ← Will be updated
     ├── progress-index.md                   ← Must be preserved & updated
-    ├── runs/
+    ├── runs/                               ← Action logs
     │   └── YYYY-MM-DD-HHMMSS-roadmap-edit/
     │       ├── agent-log.md
     │       ├── research-changes.md         ← Research on requested changes
@@ -110,7 +111,7 @@ fi
 
 LEARNING_DIR=$(grep -o '"learningDirectory"[[:space:]]*:[[:space:]]*"[^"]*"' "$OMNILEARN_CONFIG" | sed 's/"learningDirectory"[[:space:]]*:[[:space:]]*"//' | sed 's/"$//')
 OMNILEARN_DIR="$LEARNING_DIR/.omnilearn"
-SKILL_DIR="$OMNILEARN_DIR/<skill>"
+SKILL_DIR="$LEARNING_DIR/<skill>"
 ROADMAP="$SKILL_DIR/roadmap.md"
 RUNS_DIR="$SKILL_DIR/runs"
 TOPICS_DIR="$SKILL_DIR/topics"
@@ -257,7 +258,7 @@ task(category="unspecified-high", run_in_background=true, prompt="
 1. TASK: Research online to validate and inform the user's requested roadmap changes for {skill}.
 2. EXPECTED OUTCOME: A research document with web-sourced findings about the topics the user wants to add/modify.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query-docs, read, write
 
 4. MUST DO:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY research current information. Check for deprecation warnings. Prioritize resources from the last 1-2 years. Prefer latest stable versions of any library/framework/tool.
@@ -444,9 +445,9 @@ Write to `$ARCHIVE_DIR/agent-log.md`:
 │  • Preserved: All progress data intact                            │
 │                                                                   │
 │  🔗 Previous version archived at:                                 │
-│     .omnilearn/{skill}/runs/{timestamp}-roadmap-edit/             │
+│     {SKILL_DIR}/runs/{timestamp}-roadmap-edit/                    │
 │                                                                   │
-│  📍 Updated roadmap: .omnilearn/{skill}/roadmap.md                │
+│  📍 Updated roadmap: {SKILL_DIR}/roadmap.md                       │
 │                                                                   │
 │  ┌─ Change Preview ──────────────────────────────────────────┐   │
 │  │ {brief summary of the most significant changes}            │   │
@@ -463,7 +464,7 @@ Write to `$ARCHIVE_DIR/agent-log.md`:
 
 ```bash
 if git rev-parse --git-dir > /dev/null 2>&1; then
-  git add "$OMNILEARN_DIR/"
+  git add "$LEARNING_DIR/"
   git commit -m "omnilearn: update roadmap for {skill}
 
 - {change summary}
@@ -502,8 +503,8 @@ If no git repo: same flow as `/omnilearn-roadmap` — ask if the user wants to i
 | User says "you deleted my progress!" | STOP. Restore from archive. The topics/ directory must be fully intact with git checkout. |
 | Change analysis conflicts with online research | Present both perspectives, ask user which to follow |
 | Progress markers accidentally changed | Restore from archived old roadmap |
-| topic-progress.md accidentally modified | Revert: `git checkout .omnilearn/{skill}/topics/{topic}/topic-progress.md` |
-| topics/ directory accidentally modified | Revert: `git checkout .omnilearn/{skill}/topics/` |
+| topic-progress.md accidentally modified | Revert: `git checkout {SKILL_DIR}/topics/{topic}/topic-progress.md` |
+| topics/ directory accidentally modified | Revert: `git checkout {SKILL_DIR}/topics/` |
 | Topic renamed but topic-progress.md path broken | Move directory back, use a symlink or alias instead |
 | User's requested change is too vague | Ask clarifying questions before proceeding |
 

package/commands/omnilearn-roadmap.md

@@ -1,5 +1,5 @@
 ---
-description: Create a comprehensive, real-world-ready learning roadmap for any skill or subject. Uses multi-agent research to design optimal learning paths with practical, hands-on focus. Stores structured output in .omnilearn/ for continuous learning.
+description: Create a comprehensive, real-world-ready learning roadmap for any skill or subject. Uses multi-agent research to design optimal learning paths with practical, hands-on focus. Stores structured output in the learning directory for continuous learning.
 ---
 
 # /omnilearn-roadmap — Create a Comprehensive Learning Roadmap
@@ -15,13 +15,14 @@ description: Create a comprehensive, real-world-ready learning roadmap for any s
 ## Directory Structure
 
 ```
-.omnilearn/
-├── UserPreferences.md                           ← Global user preferences (auto-read + updated)
-└── <skill-name>/                                 ← Normalized skill folder (e.g., "Rust", "Machine-Learning")
-    ├── SkillPreferences.md                       ← Per-skill learning preferences
-    ├── SkillConventions.md                       ← 🔑 Setup conventions: package manager, project structure, deps, testing (auto-learned)
-    ├── roadmap.md                                ← Master roadmap (the main deliverable)
-    ├── progress-index.md                         ← Progress log — index of all runs
+.omnilearn/                                        ← ONLY config and global preferences
+├── UserPreferences.md                            ← Global user preferences (auto-read + updated)
+
+<skill-name>/                                      ← Skills at LEARNING DIRECTORY ROOT
+├── SkillPreferences.md                           ← Per-skill learning preferences
+├── SkillConventions.md                           ← 🔑 Setup conventions: package manager, project structure, deps, testing (auto-learned)
+├── roadmap.md                                    ← Master roadmap (the main deliverable)
+├── progress-index.md                             ← Progress log — index of all runs
     ├── runs/                                     ← Research & execution logs
     │   └── YYYY-MM-DD-HHMMSS-roadmap-creation/
     │       ├── agent-log.md                      ← What this run did, decisions made
@@ -87,7 +88,10 @@ OMNILEARN_DIR="$LEARNING_DIR/.omnilearn"
 echo "Using learning directory: $LEARNING_DIR"
 ```
 
-All subsequent paths in this command use `$OMNILEARN_DIR` as the base.
+**Path convention:**
+- `$OMNILEARN_DIR` = `$LEARNING_DIR/.omnilearn/` — **ONLY** config + UserPreferences.md
+- `$SKILL_DIR` = `$LEARNING_DIR/<skill>/` — skill folders are at the learning directory ROOT
+- All cross-skill scans use `$LEARNING_DIR/*/` (excluding `.omnilearn`)
 
 ## Current Date Context (CRITICAL — Must Pass to ALL Subagents)
 
@@ -102,9 +106,9 @@ CURRENT_YEAR=$(date +%Y)
 
 All subagents that use MCP tools MUST follow these exact calling conventions:
 
-### `context7_resolve-library-id` + `context7_query_docs`
+### `context7_resolve-library-id` + `context7_query-docs`
 1. **ALWAYS call `context7_resolve-library-id` FIRST** with the library name to get the correct library ID.
-2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query_docs`.
+2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query-docs`.
 3. Do NOT guess or hardcode library IDs.
 4. Max 3 calls per question.
 
@@ -128,7 +132,7 @@ All subagents that use MCP tools MUST follow these exact calling conventions:
 | `task(category="unspecified-high", background)` | Content creation | Heavy research, roadmap synthesis, file writing |
 | `task(category="deep", background)` | Autonomous multi-step | Complex subtasks that need internal orchestration |
 | `google_search` / `websearch_web_search_exa` | Research phases | Web research for content, learning paths, best practices |
-| `context7_resolve-library-id` + `context7_query_docs` | Tech skills | Official documentation for languages, frameworks, libraries |
+| `context7_resolve-library-id` + `context7_query-docs` | Tech skills | Official documentation for languages, frameworks, libraries |
 | `read`, `write`, `edit`, `bash`, `grep`, `glob` | Every phase | File operations and navigation |
 | `bash(git ...)` | Completion phase | Git commit after roadmap creation |
 
@@ -147,18 +151,18 @@ If the input is too vague (e.g., "I want to learn something"), ask for clarifica
 ### 0.2 Check Prerequisites
 
 ```bash
-SKILL_DIR="$OMNILEARN_DIR/<skill-name>"
+SKILL_DIR="$LEARNING_DIR/<skill-name>"
 RUNS_DIR="$SKILL_DIR/runs"
 TOPICS_DIR="$SKILL_DIR/topics"
 
 # Create base omnilearn dir if not exists
-mkdir -p "$OMNILEARN_DIR"
+mkdir -p "$SKILL_DIR"
 ```
 
 ### 0.3 Check for Existing Skill
 
 If `$SKILL_DIR` already has a `roadmap.md`, inform the user:
-> "A roadmap for **{skill}** already exists at `.omnilearn/{skill}/roadmap.md`. To revise it, use `/omnilearn-roadmap-edit`. To start learning, use `/omnilearn-start`."
+> "A roadmap for **{skill}** already exists at `{SKILL_DIR}/roadmap.md`. To revise it, use `/omnilearn-roadmap-edit`. To start learning, use `/omnilearn-start`."
 
 If the user wants to overwrite, note it and proceed (archive old roadmap to runs/ first).
 
@@ -182,9 +186,11 @@ Read `UserPreferences.md` if it exists. Key signals to extract:
 **This is mandatory.** Scan the learning directory for ALL existing skills the user has already learned or is learning:
 
 ```bash
-# List all existing skill directories inside .omnilearn/
-ls -d "$OMNILEARN_DIR"/*/ 2>/dev/null | while read dir; do
+# List all existing skill directories at learning root (NOT .omnilearn/)
+ls -d "$LEARNING_DIR"/*/ 2>/dev/null | while read dir; do
   skill_name=$(basename "$dir")
+  # Skip .omnilearn — it's config only
+  [ "$skill_name" = ".omnilearn" ] && continue
   # Read their progress to understand what the user already knows
   if [ -f "$dir/progress-index.md" ]; then
     echo "Existing skill: $skill_name — checking progress..."
@@ -225,7 +231,7 @@ task(category="unspecified-high", run_in_background=true, prompt="
 
 2. EXPECTED OUTCOME: A comprehensive, structured markdown file covering all knowledge areas, concepts, tools, and practices needed for real-world proficiency in {skill} — **filtered and prioritized based on what the user already knows**.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query-docs, read, write
 
 4. MUST DO:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY search for and reference current information. Prioritize resources from the last 1-2 years. Check for deprecation warnings on any library/framework/tool. Prefer latest stable versions. If something has been superseded (e.g. Create React App → Vite, npm → pnpm/uv), flag it.
@@ -367,7 +373,7 @@ task(category="unspecified-high", run_in_background=false, timeout=300000, promp
 1. TASK: Synthesize research into a personalized, level-adaptive, cross-skill-integrated master roadmap for {skill}.
 2. EXPECTED OUTCOME: A detailed roadmap.md file with a learning path TAILORED to this user's level and existing knowledge — skipping what they already know, integrating related skills, and focused on real-world readiness.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query_docs, read, write, grep
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query-docs, read, write, grep
 
 4. MUST DO:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY reference current information. If you do additional online research, prioritize resources from the last 1-2 years. Check for deprecation warnings. Prefer latest stable versions of any library/framework/tool mentioned.
@@ -535,7 +541,7 @@ Write to the agent log file documenting:
 │  ✅ Roadmap Created: {skill}                                      │
 ├──────────────────────────────────────────────────────────────────┤
 │                                                                   │
-│  📍 Location: .omnilearn/{skill}/roadmap.md                       │
+│  📍 Location: {SKILL_DIR}/roadmap.md                                │
 │                                                                   │
 │  📊 Roadmap Overview:                                             │
 │  • Foundation: {N} topics — build the fundamentals                │
@@ -582,7 +588,7 @@ fi
 
 If git repo exists:
 ```bash
-git add "$OMNILEARN_DIR/"
+git add "$LEARNING_DIR/"
 git commit -m "omnilearn: create learning roadmap for {skill}
 
 - Added master roadmap with foundation, core, advanced, and real-world topics
@@ -597,7 +603,7 @@ If no git repo exists, ask:
 If yes:
 ```bash
 git init
-git add "$OMNILEARN_DIR/"
+git add "$LEARNING_DIR/"
 git commit -m "omnilearn: initialize learning environment with {skill} roadmap"
 ```
 
@@ -613,7 +619,7 @@ Mark the task as complete. Update UserPreferences.md if new relevant information
 |-------|-------|-----------------|
 | Content research subagent completed | 1 | Block — must have research |
 | Learning path subagent completed | 1 | Block — must have research |
-| Cross-skill inventory scanned | 0 | Scan .omnilearn/ directories |
+| Cross-skill inventory scanned | 0 | Scan $LEARNING_DIR/*/ (skip .omnilearn) |
 | User level extracted from preferences | 0 | Infer from available signals |
 | Roadmap has 5+ substantive topics | 2 | Re-synthesize with fix |
 | Roadmap skips topics user already knows | 2 | Re-synthesize — remove redundant basics |

package/commands/omnilearn-start.md

@@ -27,12 +27,15 @@ The goal metric: **Can the user apply what they learned to real-world problems t
 ## Directory Structure
 
 ```
-.omnilearn/<skill>/
-├── roadmap.md
-├── SkillPreferences.md
-├── SkillConventions.md                  ← 🔑 Setup conventions: package manager, project structure, deps, testing preferences (auto-learned)
-├── progress-index.md                   ← Overview index: links to topic-progress.md per topic
-├── runs/                                   ← Skill-level run logs (action logs only, NOT progress)
+{learningDirectory}/
+├── .omnilearn/                              ← Config only (config.json, UserPreferences.md)
+│
+└── <skill>/                                  ← Skills at learning root
+    ├── roadmap.md
+    ├── SkillPreferences.md
+    ├── SkillConventions.md                  ← 🔑 Setup conventions: package manager, project structure, deps, testing (auto-learned)
+    ├── progress-index.md                   ← Overview index: links to topic-progress.md per topic
+    ├── runs/                                   ← Skill-level run logs (action logs only, NOT progress)
 │   └── YYYY-MM-DD-HHMMSS-learning-<topic>/
 │       ├── agent-log.md                    ← What happened this run: decisions, actions taken
 │       ├── interaction-1.md                ← User Q&A interaction record
@@ -71,9 +74,9 @@ CURRENT_YEAR=$(date +%Y)
 
 All subagents that use MCP tools MUST follow these exact calling conventions:
 
-### `context7_resolve-library-id` + `context7_query_docs`
+### `context7_resolve-library-id` + `context7_query-docs`
 1. **ALWAYS call `context7_resolve-library-id` FIRST** with the library name to get the correct library ID.
-2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query_docs`.
+2. Use the returned library ID (format: `/org/package`) as the `libraryId` parameter in `context7_query-docs`.
 3. Do NOT guess or hardcode library IDs.
 4. Max 3 calls per question.
 
@@ -98,7 +101,7 @@ All subagents that use MCP tools MUST follow these exact calling conventions:
 | `task(category="writing")` | Content writing | Topic explanations, solution guides |
 | `task(category="unspecified-high", ["programming"])` | Technical work | Test scripts, scaffold code |
 | `google_search` / `websearch_web_search_exa` | Research | Learning resources, topic best practices |
-| `context7_query_docs` | Tech skills | Official docs for languages/frameworks |
+| `context7_query-docs` | Tech skills | Official docs for languages/frameworks |
 | `question` tool | User interaction | Present topic choices, ask for preferences |
 | `read`, `write`, `edit`, `bash`, `grep`, `glob` | Every phase | File operations |
 | `bash(git ...)` | Completion | Commit progress |
@@ -127,7 +130,7 @@ fi
 
 LEARNING_DIR=$(grep -o '"learningDirectory"[[:space:]]*:[[:space:]]*"[^"]*"' "$OMNILEARN_CONFIG" | sed 's/"learningDirectory"[[:space:]]*:[[:space:]]*"//' | sed 's/"$//')
 OMNILEARN_DIR="$LEARNING_DIR/.omnilearn"
-SKILL_DIR="$OMNILEARN_DIR/<skill>"
+SKILL_DIR="$LEARNING_DIR/<skill>"
 ROADMAP="$SKILL_DIR/roadmap.md"
 PROGRESS_INDEX="$SKILL_DIR/progress-index.md"
 TOPICS_DIR="$SKILL_DIR/topics"
@@ -158,8 +161,9 @@ Read these files to understand the current state:
 
 **Cross-skill inventory** — Scan the learning directory for other skills the user has learned:
 ```bash
-ls -d "$OMNILEARN_DIR"/*/ 2>/dev/null | while read dir; do
+ls -d "$LEARNING_DIR"/*/ 2>/dev/null | while read dir; do
   skill_name=$(basename "$dir")
+  [ "$skill_name" = ".omnilearn" ] && continue
   if [ "$skill_name" != "$SKILL_NAME" ] && [ -f "$dir/progress-index.md" ]; then
     echo "Related skill: $skill_name — can integrate with $SKILL_NAME"
   fi
@@ -211,7 +215,7 @@ task(category="deep", run_in_background=false, timeout=600000, prompt="
 1. TASK: Create a comprehensive subtopic roadmap and initial learning materials for the topic '{topic}' within the skill '{skill}'.
 2. EXPECTED OUTCOME: A detailed topic-roadmap.md with structured learning path, AND the first assignment with question, test script, scaffold, and solution guide.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query_docs, read, write, bash, grep
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_resolve-library-id, context7_query-docs, read, write, bash, grep
 
 4. MUST DO — Step-by-Step:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY research current information. Prioritize resources from the last 1-2 years. Check for deprecation warnings. Prefer latest stable versions of any library/framework/tool. If older resources reference superseded tools (e.g. CRA → Vite, pip → uv), flag and use the current standard.
@@ -359,7 +363,7 @@ Read the assignment files for the current topic:
 │  {brief description of the assignment}                           │
 │                                                                   │
 │  Files:                                                           │
-│  • Question: .omnilearn/{skill}/topics/{topic}/assignments/      │
+│  • Question: {SKILL_DIR}/topics/{topic}/assignments/             │
 │              01-{name}/question.md                                │
 │  • Scaffold: (same directory)/scaffold/                           │
 │  • Test: (same directory)/test.{ext}                              │
@@ -401,7 +405,7 @@ task(category="unspecified-high", run_in_background=false, prompt="
 1. TASK: Create a focused micro-exercise that tests the user's understanding of '{concept}' in '{skill}'. The user is struggling with this concept during assignment '{assignment}' on topic '{topic}'.
 2. EXPECTED OUTCOME: A self-contained micro-exercise (5-15 min to solve) that isolates the specific concept and lets the user figure it out by coding.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query-docs, read, write
 
 4. MUST DO:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY research current information. Check for deprecation warnings. If a library/framework has a newer standard (e.g. CRA → Vite, pip → uv), reference the current approach.
@@ -535,7 +539,7 @@ task(category="unspecified-high", run_in_background=false, timeout=300000, promp
 1. TASK: Create the next assignment ({assignment-num}) for topic '{topic}' in skill '{skill}'.
 2. EXPECTED OUTCOME: Complete assignment with question.md, test script, scaffold, and solution-guide.md at the next difficulty level.
 
-3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query_docs, read, write
+3. REQUIRED TOOLS: google_search, websearch_web_search_exa, context7_query-docs, read, write
 
 4. MUST DO:
    - **CURRENT DATE: {CURRENT_DATE}** — ONLY use current, real-world scenarios. Avoid outdated APIs, deprecated libraries, or superseded best practices. Research what's current in the industry for this topic.
@@ -699,7 +703,7 @@ Write to `$CURRENT_RUN/agent-log.md`:
 
 ```bash
 if git rev-parse --git-dir > /dev/null 2>&1; then
-  git add "$OMNILEARN_DIR/"
+  git add "$LEARNING_DIR/"
   git commit -m "omnilearn: learning session — {skill}/{topic}
 
 - Completed assignment(s): {list}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "omnilearn-workflow",
-  "version": "1.0.2",
+  "version": "1.0.4",
   "description": "OmniLearn — AI-powered adaptive learning workflow for OpenCode. Multi-agent orchestration that creates personalized roadmaps, hands-on assignments, and tracks progress across any skill.",
   "keywords": [
     "opencode",