agentsclimarketplace

Quality review

Skill cynco-labs/ai-accounting-skills/accounting-engagement/skills/quality-review

Prove the pack before finalisation — full QC with Section A mathematical blockers. Use when quality review, QC, partner checklist, "check the accounts", or prove intent before issue.From its SKILL.md

Install
npx -y skills add cynco-labs/ai-accounting-skills --skill quality-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.8 KB, 645 tokens by cl100k_base, as published. Nobody here has run it

/quality-review

Purpose

Independent checklist pass before anything is called final. Main job: prove.

Load shared/guardrails.md and execute every item in references/qc_checklist.md (plugin or repo root shim).

Preconditions

  1. Active engagement on disk (engagement_state.json).
  2. Adjusted TB present when FS claimed — roll_tb derived, not freestyle.
  3. Re-read workpapers/FS if context was compacted (disk is truth).

Steps

1 — Section A mathematical integrity [BLOCKERS]

Run every Section A check in references/qc_checklist.md against artifacts on disk.

CheckPass when
TB DR = CRtb_adjusted.json (or claimed TB) difference 0
BS balancesAssets = liabilities + equity
P&L ↔ RERE movement ties to profit and distributions
Each JE balancesPeriod + YE journals
Bank GL = reconDiff 0.00 per bank or with limitation logged
Cash flow ↔ cashNet CF explains cash movement

Done when: every Section A item is Pass, or any Fail is written and finalisation is blocked.

2 — Sections B–E

Execute Data integrity, Standards, Completeness, Format from the same checklist. Material fails → queries or fix list; do not silent-pass.

Standards-aware classify (prove):

If classify_depth is standards_aware or engagement_type is year_end / compilation:

  • Material money-in without workpapers/analysis/revenue_recognition.mdFail C6b or documented with limitation
  • Material capex candidates without workpapers/analysis/capital_vs_expense.mdFail C6c or with limitation
  • Analysis conclusions must not contradict coded transactions.json without an open query

Done when: every checklist row has Pass / Fail / N/A with evidence path.

3 — Report + state

Write QC report under the engagement (e.g. workpapers/qc_report.md or firm path). Update engagement_state.json: prove stage result; must stop if any Section A Fail.

Done when: report on disk and state reflects pass or blocker.

Gates

  • Any Section A Fail → cannot finalise (blocker).
  • Provenance gaps on material figures → Fail Section B, not a soft skip.

Failure modes

FailureBehavior
Agent scores from memoryRe-open TB / BS / recon files; re-run checks
TB freestyled in chatReject; require roll_tb artifact
Premature “QC passed”Exhaustive bar: every checklist row statused

Output

QC report Pass/Fail per item. Next: finalise-accounts only if Section A clean (and firm policy).

Trust surface

Engagement folder + firm config. No external send.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,506. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.