# ECC functional log (host: macOS 26.5.1, aarch64, Node 24.13.1)
# Verifying the README's core claims against real output.

# Claim 1: "278 skills" -- verified via validate-skills + filesystem
$ find skills -name SKILL.md | wc -l
278

# Claim 2: "67 agents" -- verified
$ find agents -type f | wc -l
67

# Claim 3: every agent carries a Prompt Defense Baseline (anti-injection)
$ head -1 agents/a11y-architect.md && grep -c "Prompt Defense Baseline" agents/*.md | grep -v ':0' | wc -l
67/67 agents contain a "Prompt Defense Baseline" section

# Claim 4: Hermes is a first-class target
$ node scripts/install-apply.js --dry-run --json --target hermes --profile minimal | jq .plan.adapter
{ "id": "hermes-home", "target": "hermes", "kind": "home" }
$ ls docs/HERMES-SETUP.md
docs/HERMES-SETUP.md (5033 bytes)   # dedicated Hermes setup guide exists

# Claim 5: test suite is green
$ npm test | tail -7
Total Tests: 3146
Passed:      3146
Failed:         0
EXIT 0

# Claim 6: install is reversible & health-checked
$ node scripts/doctor.js --target hermes       -> Status: OK, issues: none
$ node scripts/uninstall.js --target hermes --dry-run  -> Planned removals: 455

# Claim 7: 13 cross-harness targets
$ node scripts/install-apply.js --help | grep -A2 "^Targets:"
Targets: claude, claude-project, cursor, antigravity, codex, gemini, opencode,
codebuddy, joycode, qwen, zed, hermes, openclaw, kimi  (14 listed)

# NOT verified (out of scope for hands-on host test):
# - Claude Code plugin marketplace install (/plugin install ecc@ecc) -- needs Claude Code runtime
# - GitHub App PR audits (ecc-tools) -- needs a live repo + app install
# - ECC Pro paid tier ($19/seat) -- commercial, not testable
# - The 278 individual skills' task-level outputs -- each is a prompt contract
#   executed inside a coding agent, not a runnable script. Sampled 3 for
#   structure/quality (configure-ecc, nanoclaw-repl, a11y-architect agent) only.