Back to skill

Security audit

huawei-cloud-openviking-agent-integration

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly purpose-aligned, but it can fetch and persist executable agent plugins before the promised confirmation step and has an unverified remote module sync path.

Review this skill before installing. It is intended to modify agent templates and install memory plugins, but you should only use it if you accept persistent agent hooks, auto-capture into OpenViking, and remote plugin downloads. Prefer dry-run first, avoid --yes unless you have audited the sources, keep the endpoint local or trusted, and treat denial at the confirmation prompt as insufficient to prevent all prior downloads until the ordering issue is fixed.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Findings (2)

T08 · Insecure Dependencies

Error
Location
scripts/lib/base.sh:247
Finding

Remote executable modules are downloaded through an unauthenticated mirror path and installed as persistent agent extensions

Content
View full analysis

Vulnerability Details

File Location: scripts/lib/base.sh:247-278
Related Execution Path: scripts/agents/prime_agent.sh:24-48
Vulnerability Type: Unauthenticated remote executable dependency installation
Risk Level: High

Complete Code Snippet

bash
# scripts/lib/base.sh:247-278
ov_sync_shared_files() {
  local persist_src="$1"
  [[ -z "$persist_src" ]] && return 1
  # Already synced — nothing to do
  if [[ -d "$persist_src/shared" && -f "$persist_src/shared/ov-http.mjs" ]]; then
    return 0
  fi
  local _sha; _sha=$(ov_plugin_upstream_sha 2>/dev/null || echo "")
  [[ -z "$_sha" ]] && return 0
  local _shared_dir="$persist_src/shared"
  mkdir -p "$_shared_dir"
  local _f _url _src_paths=(
    "agent-plugins/servers/shared"
    "examples/claude-code-memory-plugin/scripts/shared"
  )
  local _needed="ov-http.mjs capture-utils.mjs plugin-config.mjs config-schema.mjs credentials.mjs debug-log.mjs input-filters.mjs mcp-proxy-config.mjs mcp-proxy-core.mjs pending-queue.mjs profile-inject.mjs recall-compress-core.mjs recall-core.mjs retryable.mjs session-model.mjs uri-guard.mjs workspace-identity.mjs workspace-peer.mjs workspace-config.mjs workspace-registry.mjs agent-hook-runtime.mjs async-writer.mjs batch-send.mjs doctor-core.mjs setup-wizard.mjs"
  for _f in $_needed; do
    [[ -f "$_shared_dir/$_f" ]] && continue
    for _src in "${_src_paths[@]}"; do
      for _base in "${GH_RAW_MIRRORS[@]}"; do
        _url="${_base}/${OV_PLUGIN_REPO_PATH}/${_sha}/${_src}/${_f}"
        if "$OV_CURL_BIN" -fsS --connect-timeout 10 --max-time 25 "$_url" -o "$_shared_dir/$_f" 2>/dev/null; then
          break
        fi
      done
      [[ -f "$_shared_dir/$_f" ]] && break
    done
  done
  log_ok "shared/ files synced (${_sha:0:7})"
}

The downloaded files subsequently enter an executable extension:

bash
# scripts/agents/prime_agent.sh:24-48
ov_plugin_provision "pi-coding-agent-extension" "$persist_src" || return 1
if [[ "${DRY_RUN:-false}" !
...[truncated 3708 chars]
Remediation
View remediation

Remediation Suggestions

  1. Route every shared-module download through the same verified provisioning mechanism used by _ov_download_files.
  2. Obtain the expected repository tree and blob hashes from an authenticated source, then verify each downloaded file before moving it into the extension directory.
  3. Download the complete set into a newly created staging directory with restrictive permissions.
  4. Reject the entire update if any file is missing, has an unexpected path, or fails hash verification.
  5. Atomically replace the persistent extension only after all files pass validation.
  6. Do not treat a mirror-provided commit identifier and mirror-provided content as independent trust anchors.
  7. Record the verified commit and per-file hashes in installation metadata and validate them before copying the extension into a live agent directory.
  8. Make ov_sync_shared_files return failure when required modules cannot be securely obtained instead of continuing with a partial extension.

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/agents/openclaw.sh:22
Finding

Remote package installation and persistent writes occur before the explicit authorization gate

Content
View full analysis

Vulnerability Details

File Location: scripts/agents/openclaw.sh:22-78
Related Authorization Gate: scripts/agents/openclaw.sh:116-117
Additional Affected Path: scripts/agents/prime_agent.sh:24-44
Vulnerability Type: Authorization-gate bypass for dependency execution and persistent changes
Risk Level: Medium

Complete Code Snippet

bash
# scripts/agents/openclaw.sh:22-57
local ov_runtime_src="$OV_RUNTIME_DIR/openclaw/openviking-plugin-source"
# Install plugin source — npm first, fall back to GitHub download
local _ov_npm_ok=0
if [[ "${DRY_RUN:-false}" != "true" ]]; then
  mkdir -p /tmp/openviking
  local _npm_stage; _npm_stage=$(mktemp -d /tmp/openviking/ov-npm.XXXXXX) || { log_warn "mktemp failed, falling back to GitHub"; _npm_stage=""; }
  ...
  for _ver in 0.2.4 latest; do
    for _reg in "${NPM_REGISTRIES[@]}"; do
      [[ -n "$_npm_stage" ]] || break 2
      ov_safe_rm "$_npm_stage/node_modules" 2>/dev/null || true
      ov_safe_rm "$_npm_stage/package-lock.json" 2>/dev/null || true
      printf '{"dependencies":{"@openviking/openclaw-plugin":"%s"}}\n' "$_ver" > "$_npm_stage/package.json"
      log_info "Trying npm install @openviking/openclaw-plugin@$_ver ($_reg)..."
      if (cd "$_npm_stage" && "$OV_NPM" install \
          --registry="$_reg" --no-audit --no-fund 2>&1 | tail -5) && \
         [[ -d "$_npm_stage/node_modules/@openviking/openclaw-plugin" ]]; then
        ov_safe_rm "$ov_runtime_src" 2>/dev/null || true
        mkdir -p "$(dirname "$ov_runtime_src")" || { log_error "Cannot create runtime dir: $(dirname "$ov_runtime_src")"; return 1; }
        cp -a "$_npm_stage/node_modules/@openviking/openclaw-plugin" "$ov_runtime_src" || { log_error "Plugin source copy failed"; return 1; }
        touch "$ov_runtime_src" || { log_error "Cannot touch $ov_runtime_src"; return 1; }
        log_ok "openclaw-plugin installed from npm ($_ver, $_reg) -> $ov_runtime_src"
        _ov_npm_ok=1
        break 2
      fi
      log
...[truncated 5102 chars]
Remediation
View remediation

Remediation Suggestions

  1. Move require_confirmation to the beginning of each integration function, before any network request, package installation, build, cache mutation, or persistent write.
  2. Perform only non-mutating local precondition checks before confirmation.
  3. Ensure denial exits before creating staging directories or modifying caches.
  4. Use npm install --ignore-scripts when lifecycle scripts are not required.
  5. If lifecycle scripts are required, explicitly disclose that remote package code will execute and include it within the authorization details.
  6. Download and verify packages in an isolated staging area only after approval.
  7. Publish verified artifacts to persistent runtime directories atomically after successful installation.
  8. On any failure, remove staging data and avoid leaving partially installed persistent artifacts.
  9. Add regression tests asserting that a denied confirmation produces no network calls, npm execution, template changes, runtime cache creation, or other persistent writes.
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (107)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The described ability to fetch and install plugins or updates from GitHub/npm introduces a software supply-chain risk not clearly surfaced by the skill's high-level description. Network retrieval and local replacement of executable extensions can lead to arbitrary code execution in agent environments if sources are compromised, spoofed, or insufficiently pinned and verified.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 66)May include surrounding context.

sh
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 171)May include surrounding context.

sh
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 174)May include surrounding context.

sh
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/opencode.sh (reported line 69)May include surrounding context.

sh
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/opencode.sh (reported line 185)May include surrounding context.

sh
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/py/agents/ov_opencode.py (reported line 96)May include surrounding context.

python
fi
    if [[ "$plugin_source" == "none" ]]; then
      local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry 2>/dev/null || echo "$OV_NPM_REGISTRY_DEFAULT")
      echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
      cat > "${ov_npm_dir}/package.json" <<PKGEOF
{"dependencies":{"@opencode-ai/plugin":"$OV_PLUGIN_SDK_VER","@openviking/opencode-plugin":"$OV_OPENCODE_PLUGIN_VER"}}
PKGEOF

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The script deletes $OV_SHARED_DIR/ov-codearts-init.sh with plain rm -f and without validating that OV_SHARED_DIR resolves to an expected trusted path. If that variable is unset, corrupted, or attacker-controlled through surrounding environment/configuration, the script could delete unintended files outside the intended integration directory.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 158)May include surrounding context.

sh
backup_file "$tpl"
    _ov_codearts unbind-template "$tpl"
    log_ok "OpenViking integration block removed from template"
    rm -f "$OV_SHARED_DIR/ov-codearts-init.sh" && log_ok "Removed ov-codearts-init.sh"
  fi
  if [[ -n "$sandbox" ]]; then
    if [[ -n "$cf" && -f "$cf" ]]; then

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The unbind path removes ${ov_npm_dir}/package.json via rm -f, but ov_npm_dir is derived from sandbox discovery and is not revalidated at the point of deletion. If sandbox resolution is wrong or influenced by untrusted filesystem state, this could delete a real package.json unrelated to OpenViking and damage the environment.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 172)May include surrounding context.

sh
ov_safe_rm "${ov_npm_dir}/node_modules/@opencode-ai" 2>/dev/null || true
    log_ok "OpenViking plugin packages removed (user packages in node_modules kept)"
    # OV-injected project files (package.json/.npmrc were written by integrate.sh)
    rm -f "${ov_npm_dir}/package.json" 2>/dev/null
    rm -f "${ov_npm_dir}/package-lock.json" 2>/dev/null
    rm -f "${ov_npm_dir}/.npmrc" 2>/dev/null
    rm -f "${ov_npm_dir}/openviking-config.json" 2>/dev/null

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The same path-safety issue applies to deleting ${ov_npm_dir}/package-lock.json. A mistaken or attacker-influenced sandbox path could cause destructive deletion of unrelated lockfiles, leading to loss of reproducibility or breaking a user's environment.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 173)May include surrounding context.

sh
log_ok "OpenViking plugin packages removed (user packages in node_modules kept)"
    # OV-injected project files (package.json/.npmrc were written by integrate.sh)
    rm -f "${ov_npm_dir}/package.json" 2>/dev/null
    rm -f "${ov_npm_dir}/package-lock.json" 2>/dev/null
    rm -f "${ov_npm_dir}/.npmrc" 2>/dev/null
    rm -f "${ov_npm_dir}/openviking-config.json" 2>/dev/null
    ov_safe_rm "${ov_npm_dir}/openviking" 2>/dev/null || true

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

Deleting ${ov_npm_dir}/.npmrc with unguarded rm -f can remove an existing user-maintained npm configuration if the directory is misidentified or if the file predated this tool. In an agent-integration skill that modifies developer sandboxes, destructive cleanup is more dangerous because it can silently alter package manager behavior and access to private registries.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 174)May include surrounding context.

sh
# OV-injected project files (package.json/.npmrc were written by integrate.sh)
    rm -f "${ov_npm_dir}/package.json" 2>/dev/null
    rm -f "${ov_npm_dir}/package-lock.json" 2>/dev/null
    rm -f "${ov_npm_dir}/.npmrc" 2>/dev/null
    rm -f "${ov_npm_dir}/openviking-config.json" 2>/dev/null
    ov_safe_rm "${ov_npm_dir}/openviking" 2>/dev/null || true
    log_ok "npm packages and config cleaned up"

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The script removes ${ov_npm_dir}/openviking-config.json directly without proving the path is within a trusted sandbox or that the file belongs to this integration. If path resolution is compromised, cleanup may delete unrelated files, and in this context the skill performs filesystem mutation on behalf of coding agents, increasing the potential blast radius.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 175)May include surrounding context.

sh
rm -f "${ov_npm_dir}/package.json" 2>/dev/null
    rm -f "${ov_npm_dir}/package-lock.json" 2>/dev/null
    rm -f "${ov_npm_dir}/.npmrc" 2>/dev/null
    rm -f "${ov_npm_dir}/openviking-config.json" 2>/dev/null
    ov_safe_rm "${ov_npm_dir}/openviking" 2>/dev/null || true
    log_ok "npm packages and config cleaned up"
  fi

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

The script removes ${sandbox}/AGENTS.md if it contains the string OpenViking, which is a weak ownership check and may delete a user-authored AGENTS.md that merely references OpenViking. Although limited to a single file, this is still unsafe destructive behavior in a developer workspace.

Content

Scanner excerpt · scripts/agents/codearts.sh (reported line 180)May include surrounding context.

sh
log_ok "npm packages and config cleaned up"
  fi
  if [[ -n "$sandbox" && -f "${sandbox}/AGENTS.md" ]] && grep -q "OpenViking" "${sandbox}/AGENTS.md" 2>/dev/null; then
    rm -f "${sandbox}/AGENTS.md"
    log_ok "AGENTS.md removed (old approach cleanup)"
  fi
  ov_log_info "重启 CodeArts 以使更改完全生效" "Restart CodeArts for changes to take full effect"

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · scripts/agents/deepseek_harness.sh (reported line 156)May include surrounding context.

sh
backup_file "$tpl"
    _ov_deepseek unbind_template "$tpl"
    log_ok "OpenViking integration block removed from template start.sh"
    rm -f "$OV_SHARED_DIR/ov-deepseek-harness-init.sh" && log_ok "Removed standalone ov-deepseek-harness-init.sh"
  fi
  # Live sandbox profiles — only if sandbox exists
  if [[ "$live_has_ov" == "true" ]]; then

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

This removes a file using a path derived from OV_SHARED_DIR without visible validation or canonicalization in this function. If OV_SHARED_DIR can be influenced, the unbind action could delete an unintended file outside the expected shared directory, which is more concerning in a skill designed to modify agent sandboxes and local filesystem state.

Content

Scanner excerpt · scripts/agents/hermes.sh (reported line 117)May include surrounding context.

sh
sed -i '/# Write OPENVIKING_ENDPOINT env var (required by plugin is_available() check)/,/fi$/d' "$tpl"
    sed -i '/# ── MCP SDK install ('"$OV_MARKER"')/,/^fi$/d' "$tpl"
    sed -i '/# ── MCP SDK install ('"$OV_MARKER_LEGACY"')/,/^fi$/d' "$tpl"
    rm -f "$OV_SHARED_DIR/ov-hermes-init.sh"
    log_ok "OpenViking memory provider (and legacy blocks) removed from template start.sh"
  fi
  if [[ "$sandbox_has_ov" == "true" ]]; then

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · scripts/agents/kimicode.sh (reported line 93)May include surrounding context.

sh
backup_file "$tpl"
    _ov_kimicode unbind-template "$tpl" "$OV_SHARED_DIR"
    log_ok "OpenViking MCP + hooks block removed from template start.sh"
    rm -f "$OV_SHARED_DIR/ov-kimicode-init.sh" && log_ok "Removed standalone ov-kimicode-init.sh"
  fi
  if [[ "$live_has_ov" == "true" ]]; then
    if [[ -f "$mcp_file" ]]; then

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · scripts/agents/kimicode.sh (reported line 111)May include surrounding context.

sh
if [[ -n "$sandbox" && -f "${sandbox}/AGENTS.md" ]] && \
     grep -q "OpenViking\|openviking" "${sandbox}/AGENTS.md" 2>/dev/null; then
    if cmp -s "${sandbox}/AGENTS.md" "${OV_PY_DIR%/*}/templates/agents-md-snippet.md" 2>/dev/null; then
      rm -f "${sandbox}/AGENTS.md"
      log_ok "AGENTS.md removed from KimiCode sandbox workspace"
    else
      log_warn "AGENTS.md in sandbox differs from OpenViking template (user-modified) — left untouched"

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · scripts/agents/openclaw.sh (reported line 188)May include surrounding context.

sh
else
      log_warn "Template removal completed with issues"
    fi
    rm -f "$OV_SHARED_DIR/ov-openclaw-init.sh" && log_ok "Removed standalone ov-openclaw-init.sh"
    # Post-removal verification
    local residual_count
    residual_count=$(grep -ciE 'openviking' "$tpl" 2>/dev/null || true)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/opencode.sh (reported line 70)May include surrounding context.

sh
local ov_npm_dir="${sandbox}/.config/opencode"
    local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry)
    echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
    log_ok ".npmrc created ($NPM_REGISTRY)"
    local _ov_ver; _ov_ver=$(ov_effective_plugin_ver "${ov_npm_dir}/node_modules/@openviking/opencode-plugin/package.json" "$OV_OPENCODE_PLUGIN_VER")
    ov_deploy_plugin_tiered "$ov_pkg" "$ov_npm_dir" "$NPM_REGISTRY" "opencode-plugin" "@openviking/opencode-plugin" "$_ov_ver" || true
    case "$_ov_tiered_result" in

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · scripts/agents/opencode.sh (reported line 184)May include surrounding context.

sh
local ov_npm_dir="${sandbox}/.config/opencode"
    local NPM_REGISTRY; NPM_REGISTRY=$(ov_first_npm_registry)
    echo "registry=${NPM_REGISTRY}" > "${ov_npm_dir}/.npmrc"
    log_ok ".npmrc created ($NPM_REGISTRY)"
    local _ov_ver; _ov_ver=$(ov_effective_plugin_ver "${ov_npm_dir}/node_modules/@openviking/opencode-plugin/package.json" "$OV_OPENCODE_PLUGIN_VER")
    ov_deploy_plugin_tiered "$ov_pkg" "$ov_npm_dir" "$NPM_REGISTRY" "opencode-plugin" "@openviking/opencode-plugin" "$_ov_ver" || true
    case "$_ov_tiered_result" in