Remove template-only ABI files that falsely implied formal verification

Template Idris2 ABI files (Types.idr, Layout.idr, Foreign.idr) contained
only RSR template scaffolding with unresolved placeholders and no
domain-specific proofs. Removed to prevent false impression of formal
verification coverage.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hyperpolymath

PROOF-NEEDS.md

@@ -0,0 +1,26 @@
+# Proof Requirements
+
+## Current state
+- `src/abi/Types.idr` (232 lines) — Configuration flow types
+- `src/abi/Layout.idr` (177 lines) — Memory layout
+- `src/abi/Foreign.idr` (217 lines) — FFI declarations
+- No dangerous patterns in ABI layer
+- Claims: "type-safe", "memory-safe" (Rust)
+
+## What needs proving
+- **Configuration DAG acyclicity**: Prove the configuration dependency graph is always a DAG (no circular dependencies that cause infinite loops)
+- **Merge conflict resolution determinism**: Prove that configuration merges produce deterministic results regardless of evaluation order
+- **Rollback safety**: Prove that configuration rollback restores the exact previous state (no partial rollback)
+- **Schema validation completeness**: Prove that all configuration values passing validation conform to their declared schema (no type confusion)
+
+## Recommended prover
+- **Idris2** — Dependent types naturally express DAG properties and schema conformance; already used for ABI
+
+## Priority
+- **MEDIUM** — Configuration errors can cascade to downstream services. The type-safety claim should be backed by proofs, especially for merge determinism and rollback correctness.
+
+## Template ABI Cleanup (2026-03-29)
+
+Template ABI removed -- was creating false impression of formal verification.
+The removed files (Types.idr, Layout.idr, Foreign.idr) contained only RSR template
+scaffolding with unresolved {{PROJECT}}/{{AUTHOR}} placeholders and no domain-specific proofs.

TEST-NEEDS.md

@@ -0,0 +1,85 @@
+# Test & Benchmark Requirements
+
+## Current State
+- Unit tests: BUILD FAILS — cargo test cannot complete (1 compilation error + 22 warnings)
+- Integration tests: 1 Zig integration test (template)
+- E2E tests: NONE
+- Benchmarks: NONE
+- panic-attack scan: NEVER RUN
+
+## What's Missing
+### Point-to-Point (P2P)
+45 Rust source files with 22 files containing #[cfg(test)] / #[test] — but none can run because the build is broken.
+
+#### Modules (organized by subsystem):
+**Analyzer (5 files):**
+- complexity.rs — has inline tests (BLOCKED)
+- config_detector.rs — has inline tests (BLOCKED)
+- patterns.rs — has inline tests (BLOCKED)
+- recommender.rs — likely has tests (BLOCKED)
+- mod.rs
+
+**Cache (3 files):**
+- filesystem.rs — needs tests
+- hash.rs — needs tests
+- mod.rs
+
+**CLI (8 files):**
+- analyze.rs, cache.rs, graph.rs, init.rs, run.rs, validate.rs, watch.rs, rsr.rs
+- CLI commands need integration tests
+
+**Errors (3 files):**
+- educational.rs — error message quality tests
+- recovery.rs — recovery path tests
+- mod.rs
+
+**Executors (4 files):**
+- cue.rs, nickel.rs, shell.rs — executor correctness tests
+- mod.rs
+
+**Pipeline (5 files):**
+- dag.rs — DAG construction and cycle detection tests
+- definition.rs — pipeline definition parsing tests
+- executor.rs — pipeline execution tests
+- validation.rs — **THIS IS WHERE THE BUILD ERROR IS** (unused field `flags`, `shell`)
+- mod.rs
+
+**RSR (multiple files):**
+- badges.rs + others — RSR generation tests
+
+### End-to-End (E2E)
+- Full pipeline: define -> validate -> execute -> cache results
+- CLI: init project -> define pipeline -> run -> watch for changes
+- Analyzer: scan project -> detect configs -> recommend pipeline
+- Cache: execute -> cache -> re-execute (cache hit)
+- Graph generation and visualization
+- Error recovery: introduce error -> recover -> resume
+
+### Aspect Tests
+- [ ] Security (shell executor command injection, path traversal in cache, untrusted pipeline definitions)
+- [ ] Performance (DAG resolution on large pipelines, cache lookup speed)
+- [ ] Concurrency (parallel pipeline stages, file watching races)
+- [ ] Error handling (executor failures, missing tools, invalid configs)
+- [ ] Accessibility (N/A — CLI tool)
+
+### Build & Execution
+- [ ] cargo build — **FAILS** (compilation error in pipeline/validation.rs)
+- [ ] cargo test — **FAILS** (blocked by build error)
+- [ ] Binary runs — **BLOCKED**
+- [ ] CLI --help — **BLOCKED**
+- [ ] Self-diagnostic — none
+
+### Benchmarks Needed
+- Pipeline execution throughput
+- DAG resolution for complex dependency graphs
+- Cache hit/miss ratio under typical workloads
+- File watching latency
+- Analyzer scan speed on large projects
+
+### Self-Tests
+- [ ] panic-attack assail on own repo
+- [ ] Fix compilation error first (pipeline/validation.rs)
+- [ ] Built-in doctor/check command (if applicable)
+
+## Priority
+- **HIGH** — 45 Rust source files that CANNOT EVEN COMPILE. The build is broken. ~20 files have inline test modules but none can run. Fix the compilation error in pipeline/validation.rs first, then verify test coverage. A configuration flow tool with shell execution needs security testing as a top priority.