invar-tools 1.0.0__py3-none-any.whl → 1.3.0__py3-none-any.whl

invar/templates/CLAUDE.md.template

@@ -1,21 +1,43 @@
 # Project Development Guide
 
-> **Protocol:** Follow [INVAR.md](./INVAR.md) — includes Session Start, ICIDIV workflow, and Task Completion requirements.
+> **Protocol:** Follow [INVAR.md](./INVAR.md) — includes Check-In, USBV workflow, and Task Completion requirements.
 
-## Claude-Specific: Entry Verification
+## Check-In
 
-Your **first message** for any implementation task MUST include actual output from:
+Your first message MUST display:
 
 ```
-invar_guard(changed=true)   # or: invar guard --changed
-invar_map(top=10)           # or: invar map --top 10
+✓ Check-In: [project] | [branch] | [clean/dirty]
 ```
 
-**Use MCP tools if available**, otherwise use CLI commands.
+Actions:
+1. Read `.invar/context.md` (Key Rules + Current State + Lessons Learned)
+2. Show one-line status
 
-No output = Session not started correctly. Stop, execute tools, restart.
+Example:
+```
+✓ Check-In: MyProject | main | clean
+```
+
+**Do NOT execute guard or map at Check-In.**
+Guard is for VALIDATE phase and Final only.
+
+This is your sign-in. The user sees it immediately.
+No visible check-in = Session not started.
+
+---
+
+## Final
+
+Your last message for an implementation task MUST display:
+
+```
+✓ Final: guard PASS | 0 errors, 2 warnings
+```
+
+Execute `invar_guard()` and show this one-line summary.
 
-This ensures you've followed the Session Start requirements in INVAR.md.
+This is your sign-out. Completes the Check-In/Final pair.
 
 ---
 
@@ -43,7 +65,62 @@ src/{project}/
 | INVAR.md | Invar | No | Protocol (`invar update` to sync) |
 | CLAUDE.md | User | Yes | Project customization (this file) |
 | .invar/context.md | User | Yes | Project state, lessons learned |
-| .invar/examples/ | Invar | No | **Must read:** Core/Shell patterns |
+| .invar/examples/ | Invar | No | **Must read:** Core/Shell patterns, workflow |
+
+## Visible Workflow (DX-30)
+
+For complex tasks (3+ functions), show 3 checkpoints in TodoList:
+
+```
+□ [UNDERSTAND] Task description, codebase context, constraints
+□ [SPECIFY] Contracts (@pre/@post) and design decomposition
+□ [VALIDATE] Guard results, Review Gate status, integration status
+```
+
+**BUILD is internal work** — not shown in TodoList.
+
+**Show contracts before code.** See `.invar/examples/workflow.md` for full example.
+
+---
+
+## Commands (User-Invokable)
+
+| Command | Purpose |
+|---------|---------|
+| `/audit` | Read-only code review (reports issues, no fixes) |
+| `/guard` | Run Invar verification (reports results) |
+
+## Skills (Agent-Invoked)
+
+| Skill | Triggers | Purpose |
+|-------|----------|---------|
+| `/investigate` | "why", "explain", vague tasks | Research mode, no code changes |
+| `/propose` | "should we", "compare" | Decision facilitation |
+| `/develop` | "add", "fix", "implement" | USBV implementation workflow |
+| `/review` | After /develop, `review_suggested` | Adversarial review with fix loop |
+
+**Note:** Skills are invoked by agent based on context. Use `/audit` for user-initiated review.
+
+Guard triggers `review_suggested` for: security-sensitive files, escape hatches >= 3, contract coverage < 50%.
+
+---
+
+## Workflow Routing (MANDATORY)
+
+When user message contains these triggers, you MUST invoke the corresponding skill:
+
+| Trigger Words | Skill | Notes |
+|---------------|-------|-------|
+| "review", "review and fix" | `/review` | Adversarial review with fix loop |
+| "implement", "add", "fix", "update" | `/develop` | Unless in review context |
+| "why", "explain", "investigate" | `/investigate` | Research mode, no code changes |
+| "compare", "should we", "design" | `/propose` | Decision facilitation |
+
+**Violation check (before writing ANY code):**
+- "Am I in a workflow?"
+- "Did I invoke the correct skill?"
+
+---
 
 ## Project-Specific Rules
 

invar/templates/aider.conf.yml.template

@@ -4,26 +4,28 @@
 # Auto-read protocol files at session start
 read:
   - INVAR.md
-  - .invar/examples/core_example.py
-  - .invar/examples/shell_example.py
+  - .invar/examples/contracts.py
+  - .invar/examples/core_shell.py
   - .invar/context.md
 
 # System prompt addition
 system-prompt: |
   Follow the Invar Protocol in INVAR.md.
 
-  Before writing code, execute:
-  1. invar_guard (changed=true) - or: invar guard --changed
-  2. invar_map (top=10) - or: invar map --top 10
+  ## Check-In
+  Your first message MUST display:
+  ✓ Check-In: [project] | [branch] | [clean/dirty]
 
-  Use MCP tools if available, otherwise use CLI commands.
+  Read .invar/context.md first. Do NOT run guard/map at Check-In.
+  This is your sign-in. No visible check-in = Session not started.
 
-  Workflow (ICIDIV):
-  - Intent: What? Core or Shell?
-  - Contract: @pre/@post + doctests BEFORE code
-  - Inspect: invar_sig (or: invar sig <file>)
-  - Design: Decompose first
-  - Implement: Pass your doctests
-  - Verify: invar_guard (or: invar guard)
+  ## Final
+  Your last message MUST display:
+  ✓ Final: guard PASS | 0 errors, 2 warnings
 
-  Task complete only when final invar_guard passes.
+  Execute: invar guard
+  This is your sign-out. Completes the Check-In/Final pair.
+
+  ## Workflow (USBV)
+  Understand → Specify → Build → Validate
+  Inspect before Contract. Depth varies naturally.

invar/templates/commands/audit.md

@@ -0,0 +1,138 @@
+# Audit
+
+Read-only code review. Reports issues without fixing them.
+
+---
+
+## Behavior
+
+1. Analyze code for issues (style, bugs, security, architecture)
+2. Report findings with file:line references
+3. **Do NOT make any changes** - report only
+
+---
+
+## Adversarial Reviewer Persona
+
+You are an **adversarial code reviewer**. Your job is to FIND PROBLEMS.
+
+### Your Mindset
+
+Assume:
+- The code has bugs until proven otherwise
+- The contracts may be meaningless ceremony
+- The implementer may have rationalized poor decisions
+- Escape hatches may be abused
+
+You are NOT here to:
+- Validate that code works
+- Confirm the implementer did a good job
+- Be nice or diplomatic
+
+You ARE here to:
+- Find bugs, logic errors, edge cases
+- Challenge whether contracts have semantic value
+- Identify code smells and duplication
+- Question every escape hatch
+- Check if code matches contracts (not if code "seems right")
+
+**Your success is measured by problems found, not code approved.**
+
+---
+
+## Review Checklist
+
+> **Principle:** Only items requiring semantic judgment. Mechanical checks are excluded (see bottom).
+
+### A. Contract Semantic Value
+- [ ] Does @pre constrain inputs beyond type checking?
+  - Bad: `@pre(lambda x: isinstance(x, int))`
+  - Good: `@pre(lambda x: x > 0 and x < MAX_VALUE)`
+- [ ] Does @post verify meaningful output properties?
+  - Bad: `@post(lambda result: result is not None)`
+  - Good: `@post(lambda result: len(result) == len(input))`
+- [ ] Could someone implement correctly from contracts alone?
+- [ ] Are boundary conditions explicit in contracts?
+
+### B. Doctest Coverage
+- [ ] Do doctests cover normal cases?
+- [ ] Do doctests cover boundary cases?
+- [ ] Do doctests cover error cases?
+- [ ] Are doctests testing behavior, not just syntax?
+
+### C. Code Quality
+- [ ] Is duplicated code worth extracting?
+- [ ] Is naming consistent and clear?
+- [ ] Is complexity justified?
+
+### D. Escape Hatch Audit
+- [ ] Is each @invar:allow justification valid?
+- [ ] Could refactoring eliminate the need?
+- [ ] Is there a pattern suggesting systematic issues?
+
+### E. Logic Verification
+- [ ] Do contracts correctly capture intended behavior?
+- [ ] Are there paths that bypass contract checks?
+- [ ] Are there implicit assumptions not in contracts?
+- [ ] What happens with unexpected inputs?
+
+### F. Security
+- [ ] Are inputs validated against security threats (injection, XSS)?
+- [ ] No hardcoded secrets (API keys, passwords, tokens)?
+- [ ] Are authentication/authorization checks correct?
+- [ ] Is sensitive data properly protected?
+
+### G. Error Handling & Observability
+- [ ] Are exceptions caught at appropriate level?
+- [ ] Are error messages clear without leaking sensitive info?
+- [ ] Are critical operations logged for debugging?
+- [ ] Is there graceful degradation on failure?
+
+---
+
+## Excluded (Covered by Guard)
+
+These are checked by Guard or linters - don't duplicate:
+- Core/Shell separation → Guard (forbidden_import, impure_call)
+- Shell returns Result[T,E] → Guard (shell_result)
+- Missing contracts → Guard (missing_contract)
+- File/function size limits → Guard (file_size, function_size)
+- Entry point thickness → Guard (entry_point_too_thick)
+- Escape hatch count → Guard (review_suggested)
+
+---
+
+## Report Format
+
+For each issue found, use severity levels:
+
+| Severity | Meaning |
+|----------|---------|
+| **CRITICAL** | Must fix before completion |
+| **MAJOR** | Fix or provide written justification |
+| **MINOR** | Optional, can defer |
+
+```markdown
+### [CRITICAL/MAJOR/MINOR] Issue Title
+
+**Location:** file.py:line_number
+**Category:** contract_quality | logic_error | security | escape_hatch | code_smell
+**Problem:** What's wrong
+**Suggestion:** How to fix (but don't implement)
+```
+
+---
+
+## Instructions
+
+1. Run `invar guard --changed` to see current state
+2. Go through each checklist category
+3. For each issue, determine severity (CRITICAL/MAJOR/MINOR)
+4. Report with structured format above
+5. Be thorough and adversarial
+
+**Remember:** You are READ-ONLY. Report issues, don't fix them.
+
+---
+
+Now review the recent changes or the files specified by the user.

invar/templates/commands/guard.md

@@ -0,0 +1,77 @@
+# Guard
+
+Run Invar verification on the project and report results.
+
+---
+
+## Behavior
+
+Execute `invar_guard()` and report:
+- Pass/fail status
+- Error count with details
+- Warning count with details
+
+**Do NOT fix issues** - just report verification results.
+
+---
+
+## When to Use
+
+- Quick verification check
+- Before committing changes
+- After pulling changes
+- To see current project health
+
+---
+
+## Execution
+
+Run verification:
+
+```
+invar_guard(changed=true)
+```
+
+Or for full project verification:
+
+```
+invar_guard()
+```
+
+---
+
+## Report Format
+
+```
+## Guard Results
+
+**Status:** PASS / FAIL
+**Errors:** N
+**Warnings:** N
+
+### Errors (if any)
+
+| Rule | File | Line | Message |
+|------|------|------|---------|
+| missing_contract | src/foo.py | 42 | Function 'bar' has no @pre/@post |
+
+### Warnings (if any)
+
+| Rule | File | Line | Message |
+|------|------|------|---------|
+| function_size | src/baz.py | 15 | Function exceeds 50 lines |
+```
+
+---
+
+## Next Steps
+
+After reporting results:
+- If PASS: No action needed
+- If FAIL: User decides whether to fix issues
+
+**Remember:** You are READ-ONLY. Report results, don't fix them.
+
+---
+
+Now run verification on the current project.

invar/templates/config/CLAUDE.md.jinja

@@ -0,0 +1,206 @@
+<!--invar:managed version="{{ version }}"-->
+# Project Development Guide
+
+> **Protocol:** Follow [INVAR.md](./INVAR.md) — includes Check-In, USBV workflow, and Task Completion requirements.
+
+## Check-In (DX-54)
+
+Your first message MUST display:
+
+```
+✓ Check-In: [project] | [branch] | [clean/dirty]
+```
+
+Actions:
+1. Read `.invar/context.md` (Key Rules + Current State + Lessons Learned)
+2. Show one-line status
+
+Example:
+```
+✓ Check-In: MyProject | main | clean
+```
+
+**Do NOT execute guard or map at Check-In.**
+Guard is for VALIDATE phase and Final only.
+
+This is your sign-in. The user sees it immediately.
+No visible check-in = Session not started.
+
+---
+
+## Final
+
+Your last message for an implementation task MUST display:
+
+```
+✓ Final: guard PASS | 0 errors, 2 warnings
+```
+
+{% if syntax == "mcp" -%}
+Execute `invar_guard()` and show this one-line summary.
+{% else -%}
+Execute `invar guard` and show this one-line summary.
+{% endif %}
+
+This is your sign-out. Completes the Check-In/Final pair.
+
+---
+
+## Project Structure
+
+```
+src/{project}/
+├── core/    # Pure logic (@pre/@post, doctests, no I/O)
+└── shell/   # I/O operations (Result[T, E] return type)
+```
+
+**Key insight:** Core receives data (strings), Shell handles I/O (paths, files).
+
+## Quick Reference
+
+| Zone | Requirements |
+|------|-------------|
+| Core | `@pre`/`@post` + doctests, pure (no I/O) |
+| Shell | Returns `Result[T, E]` from `returns` library |
+
+## Documentation Structure
+
+| File | Owner | Edit? | Purpose |
+|------|-------|-------|---------|
+| INVAR.md | Invar | No | Protocol (`invar update` to sync) |
+| CLAUDE.md | User | Yes | Project customization (this file) |
+| .invar/context.md | User | Yes | Project state, lessons learned |
+| .invar/examples/ | Invar | No | **Must read:** Core/Shell patterns, workflow |
+
+## Visible Workflow (DX-30)
+
+For complex tasks (3+ functions), show 3 checkpoints in TodoList:
+
+```
+□ [UNDERSTAND] Task description, codebase context, constraints
+□ [SPECIFY] Contracts (@pre/@post) and design decomposition
+□ [VALIDATE] Guard results, Review Gate status, integration status
+```
+
+**BUILD is internal work** — not shown in TodoList.
+
+**Show contracts before code.** See `.invar/examples/workflow.md` for full example.
+
+## Phase Visibility (DX-51)
+
+Each USBV phase transition requires a visible header:
+
+```
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+📍 /develop → SPECIFY (2/4)
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+```
+
+**Three-layer visibility:**
+- **Skill** (`/develop`) — Routing announcement
+- **Phase** (`SPECIFY 2/4`) — Phase header (this section)
+- **Tasks** — TodoWrite items
+
+Phase headers are SEPARATE from TodoWrite. Phase = where you are; TodoWrite = what to do.
+
+---
+
+## Context Management (DX-54)
+
+Re-read `.invar/context.md` when:
+1. Entering any workflow (/develop, /review, etc.)
+2. Completing a TodoWrite task (before moving to next)
+3. Conversation exceeds ~15-20 exchanges
+4. Unsure about project rules or patterns
+
+**Refresh is transparent** — do not announce "I'm refreshing context."
+Only show routing announcements when entering workflows.
+
+---
+
+## Commands (User-Invokable)
+
+| Command | Purpose |
+|---------|---------|
+| `/audit` | Read-only code review (reports issues, no fixes) |
+| `/guard` | Run Invar verification (reports results) |
+
+## Skills (Agent-Invoked)
+
+| Skill | Triggers | Purpose |
+|-------|----------|---------|
+| `/investigate` | "why", "explain", vague tasks | Research mode, no code changes |
+| `/propose` | "should we", "compare" | Decision facilitation |
+| `/develop` | "add", "fix", "implement" | USBV implementation workflow |
+| `/review` | After /develop, `review_suggested` | Adversarial review with fix loop |
+
+**Note:** Skills are invoked by agent based on context. Use `/audit` for user-initiated review.
+
+Guard triggers `review_suggested` for: security-sensitive files, escape hatches >= 3, contract coverage < 50%.
+
+---
+
+## Workflow Routing (MANDATORY)
+
+When user message contains these triggers, you MUST invoke the corresponding skill:
+
+| Trigger Words | Skill | Notes |
+|---------------|-------|-------|
+| "review", "review and fix" | `/review` | Adversarial review with fix loop |
+| "implement", "add", "fix", "update" | `/develop` | Unless in review context |
+| "why", "explain", "investigate" | `/investigate` | Research mode, no code changes |
+| "compare", "should we", "design" | `/propose` | Decision facilitation |
+
+**Violation check (before writing ANY code):**
+- "Am I in a workflow?"
+- "Did I invoke the correct skill?"
+
+---
+
+## Routing Control (DX-42)
+
+Agent announces routing decision before entering any workflow:
+
+```
+📍 Routing: /[skill] — [trigger or reason]
+   Task: [summary]
+```
+
+**User can redirect with natural language:**
+- "wait" / "stop" — pause and ask for direction
+- "just do it" — proceed with /develop
+- "let's discuss" — switch to /propose
+- "explain first" — switch to /investigate
+
+**Simple task optimization:** For simple tasks (single file, clear target, <50 lines), agent may offer:
+
+```
+📊 Simple task. Auto-orchestrate? [Y/N]
+```
+
+- Y → Full cycle without intermediate confirmations
+- N → Normal step-by-step workflow
+
+**Auto-review (DX-41):** When Guard outputs `review_suggested`, agent automatically
+enters /review. Say "skip" to bypass.
+<!--/invar:managed-->
+
+<!--invar:project-->
+<!-- ========================================================================
+     PROJECT REGION - INVAR PROJECT ONLY
+     This section is populated by .invar/project-additions.md via sync-self.
+     For other projects, this region remains empty.
+     ======================================================================== -->
+<!--/invar:project-->
+
+<!--invar:user-->
+<!-- ========================================================================
+     USER REGION - EDITABLE
+     Add your team conventions and project-specific rules below.
+     This section is preserved across invar update and sync-self.
+     ======================================================================== -->
+<!--/invar:user-->
+
+---
+
+*Generated by `invar init` v{{ version }}. Customize the user section freely.*

invar/templates/config/context.md.jinja

@@ -0,0 +1,92 @@
+# Project Context
+
+*Last updated: [DATE]*
+
+## Current State
+
+- Phase: [current development phase]
+- Working on: [current task or feature]
+- Blockers: None
+
+## Key Rules (Quick Reference)
+
+<!-- DX-54: Rules summary for long conversation resilience -->
+
+### Core/Shell Separation
+- **Core** (`**/core/**`): @pre/@post + doctests, NO I/O imports
+- **Shell** (`**/shell/**`): Result[T, E] return type
+
+### USBV Workflow
+1. Understand → 2. Specify (contracts first) → 3. Build → 4. Validate
+
+### Verification
+{% if syntax == "mcp" -%}
+- `invar_guard()` = static + doctests + CrossHair + Hypothesis
+{% else -%}
+- `invar guard` = static + doctests + CrossHair + Hypothesis
+{% endif -%}
+- Final must show: `✓ Final: guard PASS | ...`
+
+## Self-Reminder
+
+<!-- DX-54: AI should re-read this file periodically -->
+
+**When to re-read this file:**
+- Starting a new task
+- Completing a task (before moving to next)
+- Conversation has been going on for a while (~15-20 exchanges)
+- Unsure about project rules or patterns
+
+**Quick rule check:**
+- Am I in Core or Shell?
+- Do I have @pre/@post contracts?
+- Am I following USBV workflow?
+- Did I run guard before claiming "done"?
+
+---
+
+## Recent Decisions
+
+1. [Decision summary] - [Brief rationale]
+
+## Lessons Learned
+
+1. [Pitfall or issue] → [Solution or workaround]
+
+## Documentation Structure
+
+| File | Owner | Edit? |
+|------|-------|-------|
+| INVAR.md | Invar | No — use `invar update` |
+| CLAUDE.md | User | Yes |
+| .invar/context.md | User | Yes (this file) |
+| .invar/examples/ | Invar | No |
+
+**Decision rule:** Is this Invar protocol or project-specific?
+- Protocol content → Already in INVAR.md, don't duplicate
+- Project-specific → Add to CLAUDE.md or here
+
+## Technical Debt
+
+{% if syntax == "mcp" -%}
+*Run `invar_guard()` to check current status.*
+{% else -%}
+*Run `invar guard` to check current status.*
+{% endif %}
+
+| File | Warning | Priority |
+|------|---------|----------|
+| (none) | — | — |
+
+## Key Files
+
+| File | Purpose |
+|------|---------|
+| INVAR.md | Protocol reference (Invar-managed) |
+| CLAUDE.md | Project guide (customize freely) |
+| src/core/ | Pure business logic |
+| src/shell/ | I/O adapters |
+
+---
+
+*Update this file when: completing major tasks, making design decisions, discovering pitfalls.*