readonly-bind-mount-patterns
暂无简介
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 13
- Forks
- 2
03 SKILL.md 正文
/readonly-bind-mount-patterns
When to Use
- Agent must read source code but must NOT be able to modify it directly
- Staging writes through an overlay (CoW) before committing to real workspace
- Docker volume mounts in agent containers:
--mount type=bind,readonly - Gate L5: path-traversal prevention via filesystem-level enforcement
Do NOT use for
- Environments where overlayfs is unavailable (some Docker-in-Docker setups)
- Replacing content-based validation (read-only mount + content scan = belt + suspenders)
Linux read-only bind mount
readonly_bind() {
local src="$1"
local dest="$2"
mkdir -p "$dest"
# Bind mount then remount read-only (two-step — single --bind,ro is unreliable)
mount --bind "$src" "$dest"
mount --remount,ro "$dest"
echo "[ro-bind] $src → $dest (read-only)"
trap "umount -l '$dest'" EXIT
}
readonly_bind /workspaces/yana-ai /sandbox/workspace
Overlayfs CoW sandbox
setup_cow_workspace() {
local real_ws="$1"
local upper="/tmp/overlay-upper"
local work="/tmp/overlay-work"
local merged="/tmp/sandbox-merged"
mkdir -p "$upper" "$work" "$merged"
mount -t overlay overlay \
-o "lowerdir=${real_ws},upperdir=${upper},workdir=${work}" \
"$merged"
echo "[overlay] CoW workspace at $merged"
echo "[overlay] real writes captured in $upper"
# After agent runs, diff what changed
find "$upper" -type f | while read -r f; do
rel="${f#$upper/}"
echo "[overlay] WRITTEN: $rel"
done
}
trap 'umount /tmp/sandbox-merged 2>/dev/null || true' EXIT
setup_cow_workspace /workspaces/yana-ai
Docker read-only bind mount
# Run agent container with workspace as read-only bind
docker run --rm \
--mount "type=bind,source=/workspaces,target=/workspace,readonly" \
--mount "type=tmpfs,target=/tmp,tmpfs-size=128m" \
--read-only \
--security-opt no-new-privileges \
--cap-drop ALL \
yamtam-agent:latest \
/bin/sh -c "cd /workspace && analyze-code.sh"
Verify mount is truly read-only
verify_readonly() {
local mount_point="$1"
if touch "${mount_point}/.ro-test-$$" 2>/dev/null; then
rm -f "${mount_point}/.ro-test-$$"
echo "[verify] FAIL: $mount_point is writable" >&2
return 1
else
echo "[verify] OK: $mount_point is read-only"
fi
}
Write audit: detect attempts to write through ro mount
# Intercept write attempts in audit log
auditctl -w /workspaces -p w -k workspace_write
# Monitor
ausearch -k workspace_write -ts recent --format csv | \
awk -F',' '{print $5, $6, $7}' | sort -u
OCI config.json bind mount (read-only)
"mounts": [
{
"destination": "/workspace",
"type": "bind",
"source": "/workspaces/yana-ai",
"options": ["rbind","ro","nosuid","nodev"]
},
{
"destination": "/tmp",
"type": "tmpfs",
"source": "tmpfs",
"options": ["nosuid","nodev","size=128m"]
}
]
Anti-Fake-Pass Checklist
❌ Single --bind,ro flag without remount → mount actually writable on some kernels
❌ Overlayfs upper dir inside ro-bound lower → writes fail with EROFS silently
❌ workdir and upperdir on different filesystems → overlayfs EXDEV at mount time
❌ No audit of upper dir after agent run → cannot confirm agent wrote nothing
❌ /tmp not isolated → agent writes temp files that reference real workspace paths
❌ rbind not used → submounts of /workspaces (nested git repos) remain writable
❌ nosuid not set on bind mount → SUID binaries in workspace can escalate
04 同仓库更多 Skills
openai--sentry--sentry
Use when the user asks to inspect Sentry issues or events, summarize recent production errors, or pull basic Sentry health data via the Sentry API; perform read-only queries with the bundled script and require `SENTRY_AUTH_TOKEN`.
ito-basket-compare
Compare Itô prediction-market baskets against a user's knowledge base, portfolio notes, financial context, watchlist, or research thesis. Use for read-only basket comparison and gap analysis without investment advice or live trading.
implementing-network-traffic-baselining
Build network traffic baselines from NetFlow/IPFIX data using Python pandas for statistical analysis, z-score anomaly detection, and hourly/daily traffic pattern profiling
detecting-evasion-techniques-in-endpoint-logs
Detects defense evasion techniques used by adversaries in endpoint logs including log tampering, timestomping, process injection, and security tool disabling. Use when investigating suspicious endpoint behavior, building detection rules for evasion tactics, or conducting threat hunting for stealthy adversary activity. Activates for requests involving evasion detection, defense evasion analysis, log tampering detection, or MITRE ATT&CK TA0005.
implementing-microsegmentation-with-guardicore
Implementing microsegmentation using Akamai Guardicore Segmentation to map application dependencies, create granular network policies, visualize east-west traffic flows, and enforce least-privilege communication between workloads across data centers and cloud.
ramdisk-tmpfs-patterns
暂无简介