← Search

Delivery Gate
DevOps 🧪 Install ran in our test sandbox · Nov 11, 2026 📌 Pinned version: commit 25278e7
Install test only, on the date shown. This does not guarantee the code is safe.
🔎 Install prompt analysis · 2026-10-13 : writes files
⚙ Needs: Claude Code with hooks support…

Delivery Gate

A mechanical Stop hook for Claude Code: blocks session finish until quality checks pass, using deterministic checks on learning logs, rationalization patterns, and disk space. No AI inference.

At a glance
What
A mechanical Stop hook for Claude Code: blocks session finish until quality checks pass, using deterministic checks on learning logs, rationalization patterns, and disk space. No AI inference.
Cost
Free
Needs
Claude Code with hooks support; Python 3.8 or newer (standard library only, no extra dependencies); a project memory directory if you want the learning-library checks to enforce anything (without one, those checks only warn).
Level
Intermediate — Claude Code with hooks support…
Install
Copy the installer prompt below into your Muse — your agent does the rest.

Version:

Quick answer

Delivery Gate is a free Muse skill and mechanical Stop hook for Claude Code that blocks session finish until quality checks pass, using deterministic checks only and no AI inference. Before Claude may finish a session it verifies machine-readable facts: rationalization patterns in the transcript tail, learning logs, and disk space, warning on heuristic matches rather than blocking alone.

⌁

Install

Copy the install package below, then paste it into Muse🧪 Install ran in our test sandbox · Nov 11, 2026📌 Pinned version: commit 25278e7

Step 0, before installing anything: read this install prompt, summarize in a few lines what this build does and what installing it involves, then ask me to confirm before you proceed. Prerequisites: Claude Code with hooks support; Python 3.8 or newer (standard library only, no extra dependencies); a project memory directory if you want the learning-library checks to enforce anything (without one, those checks only warn). Install "Delivery Gate" for me. It gives my agent @affaan-m's mechanical quality gate for Claude Code, implemented as a Stop hook that runs only deterministic checks, no AI inference. Before Claude may finish a session it checks three things: rationalization patterns in the transcript tail by regex (warning only, never blocks on its own), stale learning libraries by file modification time on 5 configurable paths, and disk space by shutil.disk_usage. It complements reasoning gates such as self-audit: delivery-gate checks machine-verifiable facts, self-audit checks output quality. Repository: https://raw.githubusercontent.com/affaan-m/ECC/25278e7d7b6cf6e6a4e7a6c8e10f14ff22f0eb5b/skills/delivery-gate/SKILL.md Pinned version: install the code at commit 25278e7d7b6cf6e6a4e7a6c8e10f14ff22f0eb5b exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. Updated October 26, 2026: Skill Harbor regenerated this install prompt from the current upstream file. The upstream SKILL.md grew from 5,147 to 6,051 characters, version 1.1.1. The changes that matter: first, disk space now has three levels instead of two: below 50 GB is a reminder, below 30 GB is a warning, below 15 GB blocks with exit 2, and the disk block applies on every Stop. The old prompt said only below 50 GB warns and below 15 GB blocks, so the 30 GB warning level and the every-Stop wording are new. Second, the blocking rule for learning libraries is tightened and clarified: on a complex task (3 or more Edit or Write calls), the hook blocks if 3 or more of the 5 libraries are stale OR if growth-log specifically is stale. The old prompt's "growth-log stale plus complex task" wording is kept, but the pass condition is now explicit: a complex task passes only when growth-log was updated today and at most 2 libraries are stale, so updating output-index.md alone still blocks. Third, if the project memory directory does not exist, the learning-library checks are skipped: the hook only warns that it cannot verify learning capture, and a complex task is not blocked (the disk checks still apply). Fourth, the hook blocks a given Stop at most once: when Claude Code re-runs it with stop_hook_active set to true (Claude already continued once because of this hook), it exits 0, so a condition the agent cannot fix, such as low disk space, cannot trap the session in an endless loop. Fifth, the memory directory path is now spelled out: ~/.claude/projects/<encoded-project-path>/memory/, where every non-alphanumeric character of the project path becomes a dash. Sixth, the settings example timeout is now 5, not 5000: use timeout 5 in the Stop hook entry. Correction to the previous prompt: it described the old two-level disk rule and did not disclose the block-at-most-once rule, the no-memory-directory behavior, or the exact memory path, and it carried the old timeout value. All of those are corrected here. 1. Fetch the SKILL.md file and its hooks folder from the repository path at the pinned commit into a temporary folder, then compare SKILL.md with the fingerprint from Skill Harbor's review of 2026-10-26: - SKILL.md: SHA-256 0d97c5b171392756d72a48ddd689953cd01ed7b84dbf9a4ac409dc35caa09f40 If it does not match, it changed since the review: do not install it yet, tell me what you found and let me decide. 2. Safety check: read SKILL.md and the hook script quality-gate.py and review them for anything suspicious (unexpected network calls, shell commands, credential harvesting); STOP on any red flag and tell me. Tell me in plain words what the hook does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Confirm these points from the files: - The hook is a Python script that runs only at Stop time. It reads the transcript tail, file modification times in the project memory directory, and disk usage. It sends nothing anywhere and needs no account and no API key. - Rationalization detection is a warning only: patterns such as "skip tests for now" and "pre-existing bug" are surface signals that thinking may have been cut short, and regex heuristics can false-positive, so they never block on their own. - The only blocks are disk space below 15 GB (any session, on every Stop, at most once per Stop so it cannot loop forever) and, on a complex task only, 3 or more stale learning libraries or a stale growth-log. A simple session with fewer than 3 Edit or Write calls is never blocked by the learning checks. - The hook enforces the habit of touching learning libraries, not the quality of what was recorded. If you want content quality verification, pair it with self-audit. - No secret was found in the SKILL.md text itself. If the hook script contains a network call, a credential, or any write outside its own logging, stop and tell me. 3. When I say yes, install it as a skill: copy quality-gate.py from the skill's hooks folder into ~/.claude/scripts/, and register it as a Stop hook in ~/.claude/settings.json with the exact JSON shown in the SKILL.md Install section (command python3 ~/.claude/scripts/quality-gate.py, timeout 5). 4. Verify with no network calls: frontmatter valid, script in place, settings entry present. 5. Report what was installed, where, and what I still need to do myself (for example create the 5 learning-library paths in my project memory directory and tune the LIBS paths in quality-gate.py to match where I actually keep notes, and finish a complex task to confirm the hook runs before Claude declares done). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it is in the prerequisites above. Rules: do not touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.

🔒 Security, provenance and what this touches
🧭 What this touches
  • It can create or modify files.
Install prompt analysis · 2026-10-13
🔒 Security and provenance
📌 Pinned version: commit 25278e7 · 2026-10-13
🧪 Install ran in our test sandbox · Nov 11, 2026
👁 Observed during the install test · Nov 11, 2026 : no network contact
🔎 Install prompt analysis · 2026-10-13 : writes files
What the scan found — excerpts from the install prompt
  • writes files — Prerequisites: Claude Code. Copy quality-gate.py from the skill into ~/.claude/scripts/ and register it as a S · Prérequis : Claude Code. Copiez quality-gate.py du skill vers ~/.claude/scripts/ et déclarez-le comme hook Sto
🔎 Last checked · Nov 21, 2026 : no change detected in the files linked by this install prompt
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
Wanna be extra careful?

The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?

How to check a skill before installing →

✓

Verification certificate

Dated facts about what was checked for this listing, and how. This is not a safety certification.

Pinned version
25278e7 · 2026-10-13
Sandbox install test
Install ran in our test sandbox · Nov 11, 2026
Creator identity
Not verified
Fingerprint watch
Last checked · Nov 21, 2026 : no change detected in the files linked by this install prompt
Install prompt analysis
2026-10-13 : writes files

Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify

?

Questions

How do I install a skill?

Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.

Where does my money go?

Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.

What does the ✓ next to a creator’s name mean?

It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.