Sendit

Audit and harden software projects to commercial release standard.

A Agent Skills skill that transforms prototype-grade codebases into production-ready products. Combines an automated scanner with a structured 4-phase execution protocol, progressive reference docs, and CI-ready JSON output.

Features

• Automated scanner — Detects TODOs, debug statements, secrets, placeholder text, empty catches, test focus directives, commented-out code, and oversized files
• Four operating modes — Quick scan, audit-only, full protocol, execute-only (decision rules)
• 10 quality gates — From design system to release artifacts, with skippability guide per project type
• Tech stack auto-detection — Stack-specific tool chains in stack audit guides
• CI integration — JSON output, severity thresholds, exit codes; GitHub Actions example
• .release-audit-ignore — Whitelist known-acceptable findings (example)
• Deploy checklist — Pre/post-deploy verification (deploy-checklist.md)
• Output templates — Consistent audit reports and sign-off (audit-templates.md)

Installation

Cursor / Claude Code / Pi

git clone https://github.com/ohms/sendit.git ~/.cursor/skills/sendit
# or
git clone https://github.com/ohms/sendit.git ~/.pi/agent/skills/sendit

Standalone scanner

python scripts/release_audit_scanner.py <project-root>

Usage

Trigger with phrases like:

• "ship it" / "make this production-ready"
• "release audit" / "quick scan"
• "harden this" / "pre-deploy check"
• "audit only" / "fix the audit findings"

Modes

Mode	Trigger	Behavior
Quick scan	"quick scan", "fast audit"	Scanner only, P0/P1 summary
Audit only	"audit only", "don't fix yet"	Full audit report, no execution
Full protocol	"ship it", "harden this"	Audit → approval → execute → verify
Execute only	"fix the findings"	Execute on existing audit

Structure

sendit/
├── SKILL.md                              # Core protocol (load on trigger)
├── scripts/
│   └── release_audit_scanner.py          # Python audit scanner
├── references/
│   ├── decision-rules.md                 # Mode selection, scope, evidence rules
│   ├── audit-templates.md                # Report, status, sign-off templates
│   ├── stack-audit-guides.md             # Per-stack tool commands
│   ├── deploy-checklist.md               # Pre/post-deploy verification
│   ├── quality-gates.md                  # 10 mandatory quality gates
│   └── execution-protocol.md             # Step-by-step execution guide
└── assets/
    ├── .release-audit-ignore.example     # Whitelist pattern examples
    └── github-actions-release-audit.yml.example

Scanner CLI

# Full scan
python scripts/release_audit_scanner.py ./my-project

# CI mode — skip slow checks
python scripts/release_audit_scanner.py ./my-project --quick --format json

# Only show blockers
python scripts/release_audit_scanner.py ./my-project --severity-threshold P0

# Exclude generated files
python scripts/release_audit_scanner.py ./my-project --exclude 'vendor/**,*.generated.*'

# Ignore whitelist file
python scripts/release_audit_scanner.py ./my-project --no-audit-ignore

Exit code 1 when P0 blockers are found — suitable for CI gates.

Progressive disclosure

SKILL.md stays focused on the core workflow. Detailed content loads on demand:

• Choosing a mode → references/decision-rules.md
• Formatting output → references/audit-templates.md
• Stack-specific tools → references/stack-audit-guides.md
• Execution details → references/execution-protocol.md
• Quality criteria → references/quality-gates.md
• Deploying → references/deploy-checklist.md

License

MIT