# Verifying Assumptions

> MANDATORY verification before/after code changes. Triggers on: modifying files, calling APIs, assuming file locations, checking dependencies, first time in codebase. ALWAYS write script to file, execute, then proceed. Never inline verification.

- **Type:** Skill
- **Install:** `agentstack add skill-andrueandersoncs-claude-skill-executable-verification-claude-skill-executable-verification`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [andrueandersoncs](https://agentstack.voostack.com/s/andrueandersoncs)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [andrueandersoncs](https://github.com/andrueandersoncs)
- **Source:** https://github.com/andrueandersoncs/claude-skill-executable-verification

## Install

```sh
agentstack add skill-andrueandersoncs-claude-skill-executable-verification-claude-skill-executable-verification
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Executable Verification

**Core rule: Every assumption must be verified by writing a script file and executing it.**

## The Pattern

```
1. Identify assumption → 2. Write script FILE → 3. Execute file → 4. Act on result
```

**Critical:** "Write a script" means creating an actual file on disk, then running that file. Never execute verification logic inline.

**Minimum per task:** Write and run at least one precondition script file (before acting) AND one postcondition script file (after acting).

## Script Examples

**Correct workflow:** Write file → Execute file

```bash
# 1. Write the script to a file
cat > verification/check-config.sh  "$VERIFY_SESSION/check-port.sh" << 'EOF'
#!/bin/bash
nc -z localhost 3000 && echo "PASS: port 3000 open" || { echo "FAIL: port 3000 closed"; exit 1; }
EOF
bash "$VERIFY_SESSION/check-port.sh"
```

## Multi-Step Tasks

Never chain actions without intermediate verification scripts:

```
Step 1: [write & run pre-script] → act → [write & run post-script]
                                                    ↓
Step 2: [write & run pre-script] → act → [write & run post-script]
```

Each step's postconditions become the next step's preconditions. Verify them with script files, not inline commands.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [andrueandersoncs](https://github.com/andrueandersoncs)
- **Source:** [andrueandersoncs/claude-skill-executable-verification](https://github.com/andrueandersoncs/claude-skill-executable-verification)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-andrueandersoncs-claude-skill-executable-verification-claude-skill-executable-verification
- Seller: https://agentstack.voostack.com/s/andrueandersoncs
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.