[HINT: Session mode inference. Returns AGENT/ASK/MANUAL with confidence.]

Infer current session mode from tool patterns.

[HINT: Tool hints. Files scanned, modified, hints added.]

[HINT: Tool catalog. action=list|help. Unified tool catalog and help.]

Unified tool catalog tool consolidating tool browsing and help operations.

📊 Output: Tool catalog or detailed tool documentation 🔧 Side Effects: None ⏱️ Typical Runtime: <1 second

[HINT: Workflow mode management. action=focus|suggest|stats. Unified workflow operations.]

Unified workflow mode management tool consolidating focus, suggestions, and usage statistics.

Actions:

• action="focus": Manage workflow modes and tool groups

• action="suggest": Get mode suggestions based on context/usage

• action="stats": View tool usage analytics and patterns

📊 Output: Mode status, suggestions, or usage statistics 🔧 Side Effects: May update tool visibility and send notifications ⏱️ Typical Runtime: <100ms

Modes (focus action):

• daily_checkin: Health + overview (9 tools, 82% reduction)

• security_review: Security-focused (12 tools, 77% reduction)

• task_management: Task tools only (10 tools, 81% reduction)

• sprint_planning: Tasks + automation + PRD (15 tools, 71% reduction)

• code_review: Testing + linting (10 tools, 81% reduction)

• development: Balanced set (25 tools, 52% reduction) [default]

• debugging: Memory + testing (17 tools, 67% reduction)

• all: Full tool access (52 tools)

[HINT: Context management. action=summarize|budget|batch. Unified context operations.]

Unified context management tool consolidating summarization, budgeting, and batch operations.

📊 Output: Context operation results (summary, budget analysis, or batch summaries) 🔧 Side Effects: None ⏱️ Typical Runtime: <10ms

[HINT: Recommendations. action=model|workflow|advisor. Unified recommendation system.]

Unified recommendation tool consolidating model selection, workflow mode suggestions, and advisor consultations.

Actions:

• action="model": Recommend optimal AI model based on task

• action="workflow": Suggest AGENT vs ASK mode based on task complexity

• action="advisor": Get wisdom from trusted advisors

📊 Output: Model recommendations, mode suggestions, or advisor wisdom 🔧 Side Effects: May log consultations (advisor action) ⏱️ Typical Runtime: <1 second

[HINT: Alignment analysis. action=todo2|prd. Unified alignment analysis tool.]

Unified alignment analysis tool consolidating Todo2 and PRD alignment.

Actions:

• action="todo2": Analyze task alignment with project goals, find misaligned tasks

• action="prd": Analyze PRD alignment with persona mapping and advisor assignments

📊 Output: Alignment scores, misaligned items, recommendations 🔧 Side Effects: Creates tasks/reports based on action ⏱️ Typical Runtime: <5 seconds

[HINT: Security. action=scan|alerts|report. Vulnerabilities, remediation.]

Unified security analysis:

• action="scan": Local pip-audit dependency scan

• action="alerts": Fetch GitHub Dependabot alerts

• action="report": Combined security report (Dependabot + pip-audit)

📊 Output: Vulnerabilities by severity, remediation recommendations 🔧 Side Effects: None (read-only)

[HINT: Config generation. action=rules|ignore|simplify. Creates IDE config files.]

Unified config generation:

• action="rules": Generate .cursor/rules/*.mdc files

• action="ignore": Generate .cursorignore/.cursorindexingignore

• action="simplify": Simplify existing rule files

📊 Output: Generated files, changes made 🔧 Side Effects: Creates/updates config files (unless dry_run=True)

[HINT: Hooks setup. action=git|patterns. Install automation hooks.]

Unified hooks setup:

• action="git": Install git hooks (pre-commit, pre-push, etc.)

• action="patterns": Install pattern triggers for file/task automation

📊 Output: Installation status, hooks configured 🔧 Side Effects: Installs hooks (unless dry_run=True)

[HINT: Prompt tracking. action=log|analyze. Track and analyze prompts.]

Unified prompt tracking:

• action="log": Log a prompt iteration (requires prompt parameter)

• action="analyze": Analyze prompt patterns over time

📊 Output: Log confirmation or iteration statistics 🔧 Side Effects: Writes to .cursor/prompt_history/ (log action)

[HINT: Health check. action=server|git|docs|dod|cicd. Status and health metrics.]

Unified health check:

• action="server": Server operational status, version

• action="git": Working copy health, uncommitted changes, sync status

• action="docs": Documentation health score, broken links

• action="dod": Definition of done validation for task completion

• action="cicd": CI/CD workflow validation, runner config

📊 Output: Health status and metrics 🔧 Side Effects: Creates tasks (docs action with create_tasks=True)

[HINT: Attribution compliance check. Verify proper attribution for all third-party components.]

Checks attribution compliance across the codebase:

• Missing attribution in file headers

• Missing entries in ATTRIBUTIONS.md

• Uncredited third-party references

• Dependency license compliance

📊 Output: Attribution score (0-100), compliance status, issues found 🔧 Side Effects: Creates compliance report and optional Todo2 tasks ⏱️ Typical Runtime: 1-3 seconds

[HINT: Report generation. action=overview|scorecard|briefing|prd. Project reports.]

Unified report generation:

• action="overview": One-page project overview for stakeholders

• action="scorecard": Health metrics scorecard with component scores

• action="briefing": Advisor wisdom summary for lowest-scoring areas

• action="prd": Product requirements document from codebase

📊 Output: Generated report in specified format 🔧 Side Effects: Creates file (if output_path specified)

[HINT: Task analysis. action=duplicates|tags|hierarchy|dependencies|parallelization. Task quality and structure.]

Unified task analysis:

• action="duplicates": Find duplicate tasks by similarity

• action="tags": Consolidate/cleanup task tags

• action="hierarchy": Analyze task structure and groupings

• action="dependencies": Analyze dependency chains, detect cycles, find critical paths

• action="parallelization": Identify tasks that can run in parallel

📊 Output: Analysis results with recommendations 🔧 Side Effects: Modifies tasks (duplicates with auto_fix, tags without dry_run)

[HINT: Testing tool. action=run|coverage|suggest|validate. Execute tests, analyze coverage, suggest test cases, or validate test structure.]

Unified testing:

• action="run": Execute test suite (pytest/unittest/ctest)

• action="coverage": Analyze test coverage with threshold

• action="suggest": Suggest test cases based on code analysis

• action="validate": Validate test organization and patterns

📊 Output: Test results, coverage analysis, test suggestions, or validation report 🔧 Side Effects: May generate coverage reports, suggestion files, or validation reports

[HINT: Linting tool. action=run|analyze. Run linter or analyze problems.]

Unified linting:

• action="run": Execute linter (ruff/flake8), optionally analyze results

• action="analyze": Analyze problems JSON with resolution hints

📊 Output: Linter results or problem analysis 🔧 Side Effects: May auto-fix issues (with fix=true)

[HINT: Memory tool. action=save|recall|search. Persist and retrieve AI discoveries.]

Unified memory management:

• action="save": Store insight with title, content, category

• action="recall": Get memories for a task_id

• action="search": Find memories by query text

Categories: debug, research, architecture, preference, insight

📊 Output: Memory operation results 🔧 Side Effects: Creates/retrieves memory files

[HINT: Task discovery. action=comments|markdown|orphans|all. Find tasks from various sources.]

Discovers tasks from:

• action="comments": TODO/FIXME in code files

• action="markdown": Task lists in *.md files

• action="orphans": Orphaned Todo2 tasks (no dependencies)

• action="all": All sources combined

📊 Output: Discovered tasks with locations 🔧 Side Effects: Can create Todo2 tasks (create_tasks=true)

[HINT: Task workflow. action=sync|approve|clarify|clarity|cleanup. Manage task lifecycle.]

Unified task workflow tool consolidating sync, approval, clarification, clarity improvement, and stale cleanup.

Actions:

• action="sync": Sync TODO markdown tables ↔ Todo2

• action="approve": Bulk approve/move tasks by status

• action="clarify": Manage task clarifications (sub_action: list|resolve|batch)

• action="clarity": Improve task clarity (adds estimates, renames, removes dependencies)

• action="cleanup": Move stale In Progress tasks back to Todo

📊 Output: Workflow operation results 🔧 Side Effects: Modifies task states ⏱️ Typical Runtime: <1 second (most actions)

[HINT: Estimation. action=estimate|analyze|stats. Unified task duration estimation tool.]

Unified task duration estimation tool consolidating estimation, accuracy analysis, and statistics.

Actions:

• action="estimate": Generate MLX-enhanced time estimate for a task

• action="analyze": Analyze estimation accuracy from historical data

• action="stats": Get statistical summary of historical task durations

📊 Output: Estimation results based on action 🔧 MLX Enhancement: When use_mlx=True, combines statistical methods with MLX semantic analysis for 30-40% better accuracy ⏱️ Typical Runtime: <1 second (estimate), <2 seconds (analyze), <1 second (stats)

[HINT: Ollama. action=status|models|generate|pull|hardware|docs|quality|summary. Unified Ollama tool.]

Unified Ollama tool consolidating integration and enhanced tools.

Actions:

• action="status": Check if Ollama server is running

• action="models": List available models

• action="generate": Generate text with Ollama

• action="pull": Download/pull a model

• action="hardware": Get hardware info and recommended settings

• action="docs": Generate code documentation

• action="quality": Analyze code quality

• action="summary": Enhance context summary

[HINT: MLX. action=status|hardware|models|generate. Unified MLX tool.]

Unified MLX tool for Apple Silicon GPU acceleration.

Actions:

• action="status": Check if MLX is available

• action="hardware": Get hardware info and recommended settings

• action="models": List recommended MLX models

• action="generate": Generate text with MLX

[HINT: Git tools. action=commits|branches|tasks|diff|graph|merge|set_branch. Unified git-inspired tools.]

Unified git-inspired task management tools.

Actions:

• action="commits": Get commit history (task_id or branch required)

• action="branches": List all branches with statistics

• action="tasks": Get all tasks in a branch

• action="diff": Compare task versions

• action="graph": Generate commit graph visualization

• action="merge": Merge tasks from one branch to another

• action="set_branch": Assign task to a branch

[HINT: Session. action=prime|handoff|prompts|assignee. Unified session management tools.]

Unified session management tool consolidating auto-primer, handoff, prompt discovery, and task assignee.

Actions:

• action="prime": Auto-prime AI context at session start

• action="handoff": End/resume sessions for multi-device coordination

• action="prompts": Find relevant prompts by mode/persona/category/keywords

• action="assignee": Manage task assignments across agents/humans/hosts

[HINT: Memory maintenance. action=health|gc|prune|consolidate|dream. Lifecycle management.]

Unified memory maintenance:

• action="health": Memory system health metrics and recommendations

• action="gc": Garbage collect stale/orphaned memories

• action="prune": Remove low-value memories based on scoring

• action="consolidate": Merge similar/duplicate memories

• action="dream": Reflect on memories with wisdom advisors

📊 Output: Maintenance results with recommendations 🔧 Side Effects: Modifies memories (gc/prune/consolidate with dry_run=False)