icdev 0.0.3__py3-none-any.whl

goals/verification_iron_law.md

@@ -0,0 +1,192 @@
+# Goal: Verification Iron Law
+
+## Description
+
+No completion claims without fresh verification evidence. Claiming work is complete without running verification is dishonesty, not efficiency.
+
+**Adapted from:** [obra/superpowers](https://github.com/obra/superpowers) verification-before-completion skill.
+
+**The iron rule:** NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE. If you haven't run the verification command in this response, you cannot claim it passes.
+
+---
+
+## When to Use
+
+**ALWAYS before:**
+- Any claim that work is "complete" or "done"
+- Any expression of satisfaction ("Great!", "All working!")
+- Committing code
+- Creating pull requests
+- Moving to the next task
+- Reporting status to the user
+- Marking a TodoWrite task as completed
+
+**This rule applies to:**
+- Exact phrases ("Done!", "Complete!")
+- Paraphrases ("All set", "Ready to go")
+- Implications ("That should work now")
+- ANY communication suggesting completion or correctness
+
+---
+
+## The Gate Function
+
+```
+BEFORE claiming any status or expressing satisfaction:
+
+1. IDENTIFY: What command proves this claim?
+2. RUN: Execute the FULL command (fresh, complete)
+3. READ: Full output, check exit code, count failures
+4. VERIFY: Does output confirm the claim?
+   - If NO → State actual status with evidence
+   - If YES → State claim WITH evidence
+5. ONLY THEN: Make the claim
+
+Skip any step = unverified claim
+```
+
+---
+
+## Verification Commands by Context
+
+### Code Changes
+```bash
+# Syntax check
+python -m py_compile path/to/file.py
+
+# Unit tests
+python -m pytest tests/ -v
+
+# Specific test
+python -m pytest tests/test_specific.py::test_name -v
+
+# Linting
+python -m ruff check path/to/file.py
+```
+
+### Dashboard Changes
+```bash
+# Start dashboard
+python tools/dashboard/app.py --host 0.0.0.0 --port 8080 &
+
+# Check it starts
+curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/login
+
+# Playwright E2E (mandatory per CLAUDE.md guardrails)
+# Navigate to every new/changed page
+# Screenshot 3 viewports (desktop 1920x1080, tablet 768x1024, mobile 375x812)
+# Check console for errors (browser_console_messages)
+# Check network for 500s (browser_network_requests)
+```
+
+### Database Changes
+```bash
+python tools/db/storage.py --health --json
+```
+
+### System Health
+```bash
+python tools/testing/health_check.py --json
+```
+
+### Security
+```bash
+python tools/security/sast_runner.py --project-dir path/
+python tools/security/secret_detector.py --project-dir path/
+```
+
+### Compliance
+```bash
+python tools/compliance/crosswalk_engine.py --project-id "sparkpilot" --coverage
+```
+
+---
+
+## Evidence Format
+
+**Good (with evidence):**
+```
+All 12 tests pass:
+$ python -m pytest tests/test_widget.py -v
+  test_create_widget PASSED
+  test_get_widget PASSED
+  ...
+  12 passed in 0.34s
+```
+
+**Bad (without evidence):**
+```
+All tests should pass now.
+```
+
+---
+
+## Common Failures
+
+| Claim | Requires | NOT Sufficient |
+|-------|----------|----------------|
+| "Tests pass" | Test output: 0 failures | Previous run, "should pass" |
+| "No console errors" | `browser_console_messages` output | "I didn't see any" |
+| "Dashboard works" | Playwright navigation + screenshot | "It should render" |
+| "Bug fixed" | Failing test → fix → passing test | "Code changed" |
+| "Build succeeds" | `py_compile` exit 0 | "Linter passed" |
+| "Subagent completed" | VCS diff shows changes | Agent reports "success" |
+| "Requirements met" | Line-by-line checklist verified | "Tests passing" |
+
+---
+
+## Red Flags — STOP
+
+If you catch yourself:
+- Using "should", "probably", "seems to"
+- Expressing satisfaction before verification
+- About to commit/push without verification
+- Trusting subagent success reports without checking
+- Relying on partial verification
+- Thinking "just this once"
+- Tired and wanting work over
+- Using different words to imply success without evidence
+
+**ALL of these mean: RUN THE VERIFICATION COMMAND.**
+
+---
+
+## Rationalization Prevention
+
+| Excuse | Reality |
+|--------|---------|
+| "Should work now" | RUN the verification |
+| "I'm confident" | Confidence ≠ evidence |
+| "Just this once" | No exceptions |
+| "Linter passed" | Linter ≠ compiler ≠ runtime |
+| "Subagent said success" | Verify independently |
+| "Partial check is enough" | Partial proves nothing |
+| "Different words so rule doesn't apply" | Spirit over letter |
+
+---
+
+## ICDEV-Specific Verification Checklist
+
+Before marking any phase or feature complete:
+
+- [ ] `py_compile` passes on all new/modified .py files
+- [ ] `pytest` passes for related tests
+- [ ] New dashboard pages verified via Playwright (3 viewports)
+- [ ] Console errors: 0
+- [ ] Network 500s: 0
+- [ ] `tools/manifest.md` updated if new tools created
+- [ ] CUI markings present on new files
+- [ ] Memory file updated with implementation details
+- [ ] Audit trail events logged for state changes
+
+---
+
+## Integration
+
+| Related Goal | Relationship |
+|-------------|-------------|
+| `tdd_workflow.md` | TDD's "verify RED/GREEN" is a subset of this |
+| `systematic_debugging.md` | Phase 4 verify-fix step uses this |
+| `subagent_review.md` | Each review stage must produce evidence |
+| `integration_testing.md` | Full test suite is the ultimate verification |
+| `build_app.md` | Post-Implementation Checklist enforces this |

goals/vsm_dashboard.md

@@ -0,0 +1,76 @@
+# CUI // SP-CTI
+
+# F3: VSM Dashboard
+
+## Purpose
+
+Track value stream metrics across the development pipeline. Compute DORA metrics (deployment frequency, lead time, change failure rate, MTTR), detect bottlenecks, and visualize end-to-end pipeline flow.
+
+## Prerequisites
+
+- `data/icdev.db` initialized with project tables
+- Pipeline events recorded (commit, build, test, deploy stages)
+
+## Workflow Steps
+
+### 1. Record Stage Event
+```bash
+python tools/analytics/vsm_engine.py --record-event --project-id "sparkpilot" --stage "build" --status "success" --duration-seconds 120 --json
+```
+**Expected output:** JSON with event ID, timestamp, stage, status, and running stage metrics.
+
+### 2. Compute DORA Metrics
+```bash
+python tools/analytics/vsm_engine.py --dora --project-id "sparkpilot" --window-days 30 --json
+```
+**Expected output:** JSON with deployment_frequency, lead_time_hours, change_failure_rate, mttr_hours, and DORA level (Elite/High/Medium/Low).
+
+### 3. Detect Bottlenecks
+```bash
+python tools/analytics/vsm_engine.py --bottlenecks --project-id "sparkpilot" --json
+```
+**Expected output:** JSON array of stages ranked by average duration, wait time, and failure rate with bottleneck flags.
+
+### 4. Get Pipeline Flow
+```bash
+python tools/analytics/vsm_engine.py --pipeline-flow --project-id "sparkpilot" --json
+```
+**Expected output:** JSON with ordered stage list, throughput per stage, WIP counts, and flow efficiency percentage.
+
+### 5. Get Trend Report
+```bash
+python tools/analytics/vsm_engine.py --trend --project-id "sparkpilot" --window-days 90 --interval weekly --json
+```
+**Expected output:** JSON array of weekly DORA snapshots showing improvement or regression per metric.
+
+## Decision Reference
+
+| Decision | Description |
+|----------|-------------|
+| D-INV-9 | DORA classification uses 2024 DORA report thresholds |
+| D-INV-10 | Stage events stored in append-only table -- immutable audit trail |
+| D-INV-11 | Bottleneck detection uses statistical outlier analysis (> 1.5x IQR) |
+| D-INV-12 | Flow efficiency = value-add time / total elapsed time per pipeline run |
+
+## Edge Cases
+
+- Fewer than 5 events returns insufficient data warning instead of DORA metrics
+- Missing stage in pipeline flow marked as "no data" rather than zero
+- Overlapping events for same stage deduplicated by latest timestamp
+- Single-stage pipeline returns flow efficiency of 100%
+
+## Tier Gating
+
+| Capability | Community | Pro |
+|------------|-----------|-----|
+| Record events | 90-day retention | Unlimited history |
+| DORA metrics | Yes | Yes |
+| Bottleneck detection | Yes | Yes |
+| Custom stage definitions | No | Yes |
+| Trend reporting | Last 30 days | Unlimited window |
+
+## Security
+
+- All stage events are append-only (NIST AU compliant)
+- Pipeline data scoped to project -- no cross-project leakage
+- CUI markings applied to exported reports

goals/writeguard.md

@@ -0,0 +1,89 @@
+# WriteGuard — AI Writing Assistant Workflow
+
+## Purpose
+
+Analyze and improve writing quality for government proposals and technical documents. Runs fully air-gapped with deterministic-first analysis pipeline.
+
+## When to Use
+
+- Analyzing draft proposal sections for readability, grammar, and tone
+- Checking RFP shall-statement coverage in proposal drafts
+- Validating win theme threading across volumes
+- Detecting cross-volume numerical inconsistencies
+- Managing reusable writing snippets and templates
+- Enforcing tenant-specific style guides with ISSO locks
+- Running batch analysis across multiple documents
+
+## Tools
+
+| Tool | Purpose |
+|------|---------|
+| `tools/writing/analysis_engine.py` | Core orchestrator — `analyze(text, mode, style_guide_id)` |
+| `tools/writing/grammar_checker.py` | Regex-based grammar error detection |
+| `tools/writing/readability_scorer.py` | Flesch-Kincaid, Gunning Fog, SMOG scoring |
+| `tools/writing/tone_profiler.py` | Keyword + qwen3 scanner tone classification |
+| `tools/writing/style_enforcer.py` | Style guide rule evaluation |
+| `tools/writing/coherence_analyzer.py` | Paragraph flow and topic drift analysis |
+| `tools/writing/plagiarism_detector.py` | RAG similarity search |
+| `tools/writing/ai_content_detector.py` | Statistical AI content detection |
+| `tools/writing/rewriter.py` | LLM-assisted text improvement |
+| `tools/writing/snippet_manager.py` | CRUD for reusable writing templates |
+| `tools/writing/style_guide_manager.py` | 5-layer cascade with ISSO locks |
+| `tools/writing/batch_analyzer.py` | Multi-document batch analysis |
+| `tools/writing/govcon_bridge.py` | GovProposal integration (read-only) |
+| `tools/writing/signal_registrar.py` | Innovation Engine bridge |
+
+## Workflow
+
+### Inline Analysis
+1. User pastes text into WriteGuard
+2. `analysis_engine.analyze()` orchestrates all checks:
+   - Grammar (deterministic regex)
+   - Readability (deterministic formulas)
+   - AI detection (deterministic statistics)
+   - Tone profiling (scanner tier)
+   - Style enforcement (cascade resolution)
+   - CUI marking check
+3. Results stored in `wg_analysis_results` + `wg_analysis_findings`
+4. Findings displayed with severity, category, suggestions
+
+### Batch Analysis
+1. Submit multiple documents via `batch_analyzer.run_batch()`
+2. Each document analyzed individually
+3. Cross-document consistency checks run
+4. Results aggregated in `wg_batch_runs`
+
+### GovProposal Integration
+1. `check_shall_compliance(text, opportunity_id)` — reads `rfp_shall_statements`, checks keyword coverage
+2. `validate_win_themes(text, opportunity_id)` — reads `proposal_knowledge_base` differentiators
+3. `check_cross_volume_consistency(opportunity_id)` — reads all drafts, detects numerical contradictions
+
+### Style Guide Management
+1. Create guides at platform/tenant/program/project/user scope
+2. ISSO can lock dimensions (e.g., readability rules)
+3. Cascade resolves bottom-up, locked dimensions skip-propagate
+4. Version history with diff and rollback
+
+## Args
+
+- `args/writeguard_config.yaml` — analysis defaults, thresholds, LLM settings
+
+## Context
+
+- `context/writing/style_guides/*.yaml` — 5 default guides
+- `context/writing/grammar_rules/*.json` — grammar patterns and govcon vocabulary
+
+## Edge Cases
+
+- Empty text returns early with status message
+- Missing style guide uses platform defaults
+- Nonexistent opportunity_id returns note (not error)
+- Single-paragraph text skips connectivity/topic-drift analysis
+- Air-gapped mode: all deterministic checks work; LLM-assisted checks (tone, coherence, rewrite) degrade to deterministic-only
+
+## Security
+
+- All analysis results are append-only (NIST AU compliant)
+- GovProposal bridge is read-only — never writes to proposal tables
+- CUI marking enforcement is configurable
+- Style guide locks prevent unauthorized changes to compliance dimensions