Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Close Approval Workflow

ASecurity

The governed review→approve→lock state machine with enforced segregation of duties and an append-only hash-chained audit log. Refuses same-actor approval above threshold and illegal transitions. Runs scripts/close_state.py. Honest tier caveats. Used by `controller` + `audit-prep-specialist`.

7 stars
0 votes
0 copies
0 views
Added 9/23/2026
ai-agentspythongoshell

Security Analysis

A100/100

Scanned 9/23/2026

$npx -y skills add mcorbett51090/RavenClaude --skill close-approval-workflow --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Close Approval Workflow?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Close Approval Workflow
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/mcorbett51090-close-approval-workflow/badge)](https://www.skillsdirectory.com/skills/mcorbett51090-close-approval-workflow)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: close-approval-workflow
description: "The governed review→approve→lock state machine with enforced segregation of duties and an append-only hash-chained audit log. Refuses same-actor approval above threshold and illegal transitions. Runs scripts/close_state.py. Honest tier caveats. Used by `controller` + `audit-prep-specialist`."
---

# Skill: close-approval-workflow

**Purpose:** Make the close's controls *enforced by construction*, not documented. Every deliverable is submitted into a state machine that refuses illegal transitions and refuses a same-actor approval above the entity's SoD threshold, writing every event to an append-only, hash-chained audit log.

Engine: [`../../scripts/close_state.py`](../../scripts/close_state.py) (stdlib only).

## Why it exists

The failure mode behind the public "AP agent released ~$92K past a documented control" incident (2025) is a control that was *written down but not enforced*. Here the enforcement is code: the state machine is the only path a package can advance, and the SoD check is a hard refusal, not a policy note.

## State model

```
draft → submitted → in_review → approved → locked
                  ↘ (reject) → draft        ↘ (authorized reopen) → draft
```

- `submit` records the preparer + package amount.
- `approve` **refuses** when `actor == preparer` and `amount ≥ sod_threshold` (logged denial).
- `lock` needs a **non-agent approval token**; without it the package is badged **self-certified, single-actor** (a stderr warning + a `state.json` flag).
- `verify` recomputes the hash chain end-to-end; any retroactive edit breaks it (exit non-zero).

## Invocation

```shell
D=.ravenclaude/runs/close/<entity>/<period>
python3 scripts/close_state.py --run-dir $D submit  --actor autopilot --amount 1852500
python3 scripts/close_state.py --run-dir $D review  --actor controller
python3 scripts/close_state.py --run-dir $D approve --actor controller --threshold 100000
python3 scripts/close_state.py --run-dir $D lock    --actor controller --approval-token <out-of-band>
python3 scripts/close_state.py --run-dir $D verify
```

## HONEST TIER CAVEAT (do not overclaim)

Identity here is **config-asserted, not authenticated** — `--actor NAME` is a caller-supplied string with no identity-provider binding. The hash chain is therefore tamper-**evident** (a third party detects an edit), **not** tamper-**preventing** against the operator who holds the file. At this local tier the SoD check stops the *same declared actor* self-approving above threshold and stops the accidental single-process auto-approve — it does **not**, by itself, make the close auditor-grade, and it does **not** "design out" the Ramp incident. Real auditor-reliable segregation needs an identity provider + an immutable store, which lands at the warehouse/ELT tier. LOCK's non-agent approval-token requirement is the bridge: absent it, the package is explicitly self-certified. State this caveat wherever the workflow's controls are described.

## Orchestrator boundary

The close orchestrator ([`run-controller-cycle`](../../commands/run-controller-cycle.md)) calls **`submit` only** — reaching `approved`/`locked` always requires a separate, later, human-invoked call. This is regression-tested ([`test_controller_autopilot.py`](../../scripts/test_controller_autopilot.py)) so an unattended run can never collapse the review gate.

Attribution

mcorbett51090mcorbett51090
View sourceSee grades on GitHubMore from mcorbett51090 →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698461 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →