From midnight-verify
Run a curated set of verification claims against the current toolchain to detect behavioral changes. Each claim is verified through the normal verification pipeline (classify → dispatch agent → verify). Supports full sweep (all categories) and targeted sweep (single category). Invocable as /midnight-verify:zkir-regression or loadable as a sense-check when toolchain issues are suspected.
How this skill is triggered — by the user, by Claude, or both
Slash command
/midnight-verify:zkir-regressionThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Run a curated set of verification claims against the current toolchain to detect behavioral changes. Each claim is verified through the full pipeline — claim classification, contract writing, compilation, execution, and PLONK proof verification.
Run a curated set of verification claims against the current toolchain to detect behavioral changes. Each claim is verified through the full pipeline — claim classification, contract writing, compilation, execution, and PLONK proof verification.
Use this when:
If $ARGUMENTS is empty → full sweep (all categories).
If $ARGUMENTS contains a category name → targeted sweep (that category only).
Valid categories: arithmetic, types, state, privacy, zk-proof, transcript.
compact compile --language-version
compact --version
Record both for the report header.
For each claim in the list below (filtered by category if targeted):
midnight-verify:verify-correctness skill to classify the claim domainThis follows the same flow as the /midnight-verify:verify command — you are the orchestrator for each claim.
| ID | Category | Claim | Expected Verdict |
|---|---|---|---|
| arith-1 | arithmetic | A pure circuit that adds two Uint32 values (3 + 4) returns the correct sum (7) | Confirmed (tested) |
| arith-2 | arithmetic | A pure circuit that multiplies two Uint32 values (3 * 5) returns the correct product (15) | Confirmed (tested) |
| types-1 | types | Assigning a value of 256 to a Uint8 variable produces a compiler error | Confirmed (tested) |
| types-2 | types | A pure circuit returning a tuple allows 0-indexed access to each element | Confirmed (tested) |
| state-1 | state | A counter contract's increment circuit updates the ledger state by the specified amount | Confirmed (tested) |
| state-2 | state | Reading a counter ledger value returns the current on-chain state | Confirmed (tested) |
| privacy-1 | privacy | A circuit that writes to the ledger requires a disclose() call | Confirmed (tested) |
| zk-1 | zk-proof | A counter contract's increment circuit passes the full PLONK proof verification | Confirmed (zkir-checked) |
| zk-2 | zk-proof | Tampering with the public transcript of a verified circuit causes PLONK checker rejection | Confirmed (zkir-checked) |
| zk-3 | zk-proof | The PLONK checker error for a tampered transcript identifies the exact mismatched input | Confirmed (zkir-checked) |
| transcript-1 | transcript | A counter increment circuit encodes ledger operations in the publicTranscript | Confirmed (zkir-inspected) |
| transcript-2 | transcript | The compiled ZKIR for a counter increment contains declare_pub_input instructions | Confirmed (zkir-inspected) |
For each claim, compare the actual verdict against the expected verdict:
## ZKIR Regression Report
**Toolchain:** compact CLI vX.Y.Z, language version A.B.C
**Date:** YYYY-MM-DD
**Mode:** [full sweep / targeted: <category>]
**Ran:** N claims
### Results
| Category | Passed | Failed | Total |
|---|---|---|---|
| arithmetic | N | N | N |
| types | N | N | N |
| state | N | N | N |
| privacy | N | N | N |
| zk-proof | N | N | N |
| transcript | N | N | N |
| **Total** | **N** | **N** | **N** |
### Failures (if any)
**<claim-id>:** Expected <expected verdict>, got <actual verdict>
- Claim: "<claim text>"
- Actual result: [what the verification pipeline returned]
- Interpretation: [what this failure suggests about toolchain changes]
If there are zero failures, end with:
All N claims passed. Toolchain behavior matches expectations.
Add a row to the claim list table above. Each claim should:
/midnight-verify:verify pipelinenpx claudepluginhub devrelaicom/midnight-expert --plugin midnight-verifyHub skill for the midnight-verify plugin. Classifies claims by domain, routes to the appropriate domain skill, dispatches sub-agents based on the domain skill's routing, and synthesizes final verdicts. Loaded by the /midnight-verify:verify command — the main thread acts as orchestrator.
Enforces evidence-based completion claims by requiring fresh command output before reporting success. Use when finishing any task, fixing a bug, running tests, building, deploying, or making any "it works" claim.
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.