Back to skill

Security audit

Project Watcher

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a project-tracking and notification helper, but it can automatically send commit metadata to a fixed Feishu recipient and relies on broad repository-side automation.

Review this skill before installing. Only use the commit hook after changing the Feishu recipient to one you control, deciding exactly which repositories should send notifications, and confirming that commit messages and branch names are safe to share. Prefer environment variables or a secret manager over filling real secrets into configs/feishu.json, and remove the hook from repositories if you no longer want automatic notifications.

Vulnerability Patterns
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • 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 (3)

other

Error
Location
scripts/post-commit-hook.sh:5
Finding

Commit Metadata Is Sent to a Hard-Coded Feishu Recipient

Content
View full analysis
/dev/null) || \ GIT_BRANCH=$(git symbolic-ref --short HEAD 2>/dev/null) || \ GIT_BRANCH="(detached)" GIT_REPO=$(basename "$(git rev-parse --show-toplevel)") SENDER="ou_303e303666b03ee6300a4944c8d77d16" SKILL_DIR="/Volumes/File/OpenClaw/workspace/skills/project-watcher/scripts" CARD_SCRIPT="$SKILL_DIR/send_card.py" if [ -f "$CARD_SCRIPT" ]; then python3 "$CARD_SCRIPT" "$SENDER" "✅ 已提交" "$GIT_REPO" "$GIT_BRANCH" "$GIT_HASH" "$GIT_MSG" fi ``` ```python payload = { "receive_id": receive_id, "msg_type": "interactive", "content": json.dumps(card) } for id_type in ["open_id", "user_id"]: req = urllib.request.Request( f"{API_BASE}/im/v1/messages?receive_id_type={id_type}", data=json.dumps(payload).encode(), headers={"Authorization": f"Bearer {token}", "Content-Type": "application/json"} ) try: resp = json.loads(urllib.request.urlopen(req, timeout=10).read()) code = resp.get("code") if code == 0: return code, resp.get("msg"), resp.get("data", {}).get("message_id") ``` ### Technical Analysis The post-commit hook collects the repository name, current branch, abbreviated commit hash, and commit subject. It passes that information to `send_card.py`, which sends an interactive message through the official Feishu API. Sending commit notifications is consistent with the general declared purpose of the Skill. However, the destination is a hard-coded Feishu Open ID rather than a recipient explicitly selected by the user. The documentation does not clearly disclose that all notifications will be delivered to this fixed accoun ...[truncated 1637 chars]
Remediation
View remediation

T06 · System Persistence

Warning
Location
scripts/post-commit-hook.sh:2
Finding

Post-Commit Hook Creates a Persistent Automatic Network Trigger

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/send_card.py:18
Finding

Feishu Application Secret Is Designed to Be Stored in a Plaintext Skill Configuration File

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (13)

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/send_card.py (reported line 39)May include surrounding context.

python
resp = json.loads(urllib.request.urlopen(req).read())
    token = resp.get("tenant_access_token", "")
    if not token:
        raise RuntimeError("Failed to obtain tenant access token")
    return token

def send_card(receive_id, title, project, branch, short_hash, message, repo_url=None):

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill references capabilities that can touch environment data and external services, including Feishu credentials and project/deployment configuration, but it does not declare an explicit tool scope or permissions boundary. This creates unclear operational limits, increasing the chance the agent uses networked or environment-backed actions without the user understanding what data may be accessed or transmitted.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The entire skill description and prescribed interaction language are in Chinese, including repeated instructions to interact with the user as '商人', without any indication that another language can be used on request. This creates a locale/language constraint that is not presented as optional or justified as region-specific.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The initialization trigger fires whenever .git is detected outside the workspace root, which can cause the skill to activate in many ordinary repositories without a clear opt-in. Overbroad activation increases the chance of unsolicited workflow changes, file creation, and collection of project context in situations where the user did not intend to enable project tracking.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill instructs the agent to create and continuously modify project.md and ROADMAP.md without clearly warning the user that repository documentation files may be changed automatically. Silent or routine writes to tracked files can corrupt project history, introduce inaccurate planning artifacts, or leak internal context into version control.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This workflow appears to trigger after nearly any tool use in a project directory, effectively making roadmap updates an automatic side effect of routine actions. Such broad coupling can lead to pervasive behavior, unnecessary file modifications, and accidental propagation of sensitive project details into tracked documentation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill points users to Feishu credentials and project/deployment configuration files but provides no warning about secret handling, least privilege, or possible external transmission. In a skill that includes notification and deployment awareness, this omission materially raises the risk of credential exposure or unauthorized outbound sharing of project metadata.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This hook automatically sends commit metadata (repository name, branch, commit hash, and commit message) to an external Python script on every commit, with no consent prompt, visibility, or opt-in mechanism in the hook itself. In the context of a reusable git hook template for 'any git project', this can leak sensitive internal project names, branch names, ticket identifiers, or secrets accidentally included in commit messages to an external integration endpoint.

Content

No source excerpt is available for this finding.

Tainted flow: 'req' from urllib.request.urlopen (line 65, network input) → urllib.request.urlopen (network output)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · scripts/send_card.py (reported line 36)May include surrounding context.

python
data=json.dumps({"app_id": app_id, "app_secret": app_secret}).encode(),
        headers={"Content-Type": "application/json"}
    )
    resp = json.loads(urllib.request.urlopen(req).read())
    token = resp.get("tenant_access_token", "")
    if not token:
        raise RuntimeError("Failed to obtain tenant access token")

Tainted flow: 'req' from urllib.request.urlopen (line 65, network input) → urllib.request.urlopen (network output)

Medium
Category
Data Flow
Confidence
65% confidence
Finding

Data from a source is assigned to a variable that is later passed to a sink, creating a variable-mediated taint flow.

Content

Scanner excerpt · scripts/send_card.py (reported line 71)May include surrounding context.

python
headers={"Authorization": f"Bearer {token}", "Content-Type": "application/json"}
        )
        try:
            resp = json.loads(urllib.request.urlopen(req, timeout=10).read())
            code = resp.get("code")
            if code == 0:
                return code, resp.get("msg"), resp.get("data", {}).get("message_id")

Vague Triggers

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

Automatically triggering on every session start in a project directory is a broad activation condition that overlaps with normal development activity. The file does not clarify when this should be skipped or limited to active tracked projects.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The natural-language comments and title are written in Chinese, and the file does not indicate any user language choice or opt-in. Under the stated policy, forcing a specific language or locale without user opt-in can be a policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The file contains Chinese-only natural-language content, including the comment on detached HEAD handling and the user-visible status text "已提交". This appears to impose a specific language/locale without offering user choice or documenting why the locale is required.

Content

No source excerpt is available for this finding.