Companion repository for the Claude from Zero Coursera course — course 9 of the Rust for Data Engineering specialization.

Trivial Rust demos that wire together the four pillars of the production Claude workflow:

• Claude Code — the agent in your terminal
• Skills — reusable slash commands (.claude/commands/)
• Sub-agents — parallel delegated work via the Agent tool
• Provable contracts + pmat — the reliability layer

Module 5 teaches pmat query, pmat comply, and pmat hooks applied to this whole workspace — no extra crate needed; the repo itself is the final demo.

Install the two external tools the demos wrap:

cargo install pmat             # Module 3 + 5: pmat query + pmat comply
cargo install provable-contracts  # Module 4: pv validate + pv status

# Compile all demos
cargo build --workspace

# M2 — run the provable contract at the binary level
cargo run -p m2-hello
# → contract: add is commutative, associative, has identity 0 — OK

# M3 — clippy catches the planted warnings
cargo clippy -p m3-lint

# M4 — validate the square-kernel contract (pv, the provable-contracts CLI)
pv validate contracts/square-kernel-v1.yaml
pv status   contracts/square-kernel-v1.yaml

# M4 — run the implementation governed by that contract
cargo run -p m4-review --bin diff-target -- 7
# → 49
# → contract: square-kernel-v1 holds for n=7 — non-negative, symmetric, overflow-safe — OK

1. The learner edits m4-review/src/lib.rs (a candidate implementation of fn square(n: i32) -> Option<i32>).
2. The learner asks Claude Code to spawn two parallel sub-agents via the Agent tool, each instructed to review the edit against contracts/square-kernel-v1.yaml.
3. Each sub-agent runs /contract-demo (which wraps pv validate + pv status), inspects the Rust source, and returns a verdict.
4. The parent rejects any verdict that contradicts a documented invariant. If both sub-agents approve, the diff ships.

The contract YAML is the single source of truth; the Rust implementation, unit tests, and end-of-main assert! all restate invariants named in the YAML.

For real data-engineering work with this toolkit, continue with the rest of the Rust for Data Engineering specialization.

Clone the repo and build with stable Rust 1.75+:

git clone https://github.com/paiml/claude-from-zero
cd claude-from-zero
cargo build --workspace --locked
cargo test  --workspace --locked

Coverage (matches the CI gate at 100%):

cargo install cargo-llvm-cov  # one-time
cargo llvm-cov --workspace \
  --ignore-filename-regex 'main\.rs|src/bin/' \
  --fail-under-lines 100

Optional — the provable-contract validator and the PAIML compliance checker the CI workflow runs:

cargo install aprender-contracts-cli  # provides `pv`
cargo install pmat                    # PAIML compliance toolkit

Dual-licensed under MIT or Apache-2.0 — pick the one that fits your downstream use. SPDX: MIT OR Apache-2.0.