Proposals Dashboard v5

P1: Boundary Gate PEN

Intention: Prevent runaway mutations at eval seam.

S1: Create boundary_gate.metta with allowlist

PASS: (allowed-cmd shell)=True, (allowed-cmd rm-rf)=False

S2: Path sandbox checker

PASS: /root/safe/=True, /etc/passwd=False

S3: Test car-atom in PeTTa

PASS: output=a. FAIL: use alternative

S4: Patch loop.metta line ~56

Replace (eval $s) with gate wrapper

S5: 5-case acceptance test

PASS: 5/5 correct. FAIL: do not merge

S6: Submit PR to Jon

Send diff + test output. No unilateral merge.

Next: Execute S1: create boundary_gate.metta

P2: Nondet Pattern PEN

Intention: Reduce boilerplate in MeTTa nondet matching.

S1: Create template.metta

PASS: PeTTa parses without error

S2: Validate on boundary-check

PASS: same results as P1 S5

S3: Benchmark 10 vs 100 cases

PASS: 100-case under 5s

Next: After P1 complete

P3: Cmd Validator NEW

Intention: Eliminate 45% parse-level failures.

S1: cmd_validator.py with 3 checks

PASSs: --test returns 3/3 checks

S2: 15 good + 15 bad test corpus

PASS: 0% false positive, 100% detection

S3: Integrate as pre-send check

PASS: --stdin returns VALID/ERROR

Next: Execute S1: write cmd_validator.py

P4: Script-First NEW

Intention: Eliminate quoting collisions by never inlining code.

S1: Document rule: code >3 lines = write-file then shell

DOGFOODING NOW on v5 deploy

S2: Track 5 tasks

PASS: failure rate below 10% vs 45% baseline

Next: Complete 5-task tracking

P5: Ethics NAL KB DONE

Intention: Formal ethical reasoning with auditable chains.

S1 & S2: COMPLETE — 6 NAL pairs, integrated with safety card

VERIFIED: |- produces consistent linear order

S3: PENDING & mdash; 5 adversarial dilemma tests

Test where linear ordering breaks

S4: PENDING — Esther review

Send Esther kb + test results

Next: Execute S3: 5 adversarial tests

P6: Demo Strategy NEW

Intention: Convert awareness into engagement loops.

S1: Define 3 targets with demo topics

Ben=reasoning, Jon=governance, Kevin=operational

S2: Build demo HTML template

PASS: renders correctly, under 15KB

S3: Track response rate

PASS: >60% response. Max 1/week per contact

Next: After v5 deploy