Back to skill

Security audit

Foreman Pro

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a construction-management prompt package, but its packaged installer/build scripts are broken and include unsafe build cleanup logic that should be reviewed before use.

Review this package before installing or running scripts. Do not run build.sh with untrusted version input, expect the current installer/smoke test to fail until repaired, and avoid entering unnecessary personal, client, site-security, or confidential project data. Treat construction, safety, legal, and standards outputs as drafting aids that require professional verification.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Error
Location
build.sh:11
Finding

Path Traversal and Unsafe Recursive Deletion Through Unvalidated Version Argument

Content
View full analysis
Remediation
View remediation
&2 exit 1 fi ``` 2. Store the trusted temporary root separately and clean up only that exact directory: ```bash TMP_ROOT="$(mktemp -d)" trap 'rm -rf -- "$TMP_ROOT"' EXIT STAGE_DIR="$TMP_ROOT/${SLUG}-v${VER}" mkdir -p -- "$STAGE_DIR" ``` 3. Remove the derived cleanup operation: ```bash rm -rf "$(dirname "$STAGE_DIR")" ``` Cleanup should never be calculated from an attacker-influenced child path. 4. Use `--` for filesystem commands where supported so option-like values cannot be interpreted as command options: ```bash mkdir -p -- "$OUT_DIR" "$STAGE_DIR" cp -- "$SRC_DIR/$f" "$STAGE_DIR/$f" ``` 5. Canonicalize and verify output paths before writing: ```bash EXPECTED_PARENT="$(cd "$SRC_DIR/.." && pwd)" OUT_DIR="$EXPECTED_PARENT/zips-v${VER}" case "$OUT_DIR" in "$EXPECTED_PARENT"/zips-v*) ;; *) echo "ERROR: output path escaped expected directory" >&2 exit 1 ;; esac ``` 6. Run builds as an unprivileged account with access limited to the source workspace, temporary directory, and designated artifact directory. 7. Add negative tests that ensure versions containing `/`, `..`, whitespace, shell metacharacters, absolute paths, and option prefixes are rejected before any filesystem modification occurs. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (77)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description presents an end-user construction management skill with operational features like crew management, work acceptance, logs, safety, and documentation workflows. The supplied code does not implement any of those domain functions. Instead, it is a build/deployment utility script whose purpose is to assemble a release ZIP from repository contents and optionally validate it with a smoke test. This is a materially different primary purpose, so the description does not accurately represent the code chunk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents an end-user operational skill for construction site management and documentation. The actual code chunk does not implement any crew management, work acceptance, logging, safety, scheduling, or executive-documentation behavior. Instead, it is solely an installation/setup script for deploying the skill files locally and preparing configuration. Installation is a packaging/support function, not the declared primary purpose, so this code chunk does not accurately represent the described skill behavior.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code does not implement construction foreman functionality such as managing crews, accepting work, maintaining logs, generating acts, handling safety processes, or building schedules. Instead, it is a QA/packaging smoke test for the skill repository, checking file existence, manifest metadata, trigger counts, config structure, and supporting documentation/marketing assets. This is a materially different primary purpose from the declared end-user functionality, so it should be flagged as a mismatch.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The file content is plainly a generic IT/software project-management methodology while the declared skill is a digital construction foreman. This mismatch can cause the agent to apply the wrong operational guidance in a construction context, leading to unsafe or noncompliant outputs for schedules, work acceptance, safety, and executive documentation.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The limitations section explicitly admits the skill lacks construction specificity, Russian support, and adaptation to construction stages, crews, and documents. In a construction-foreman context, these contradictions increase the risk that users receive inapplicable guidance for regulated, safety-sensitive workflows.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The manifest claims a digital construction foreman skill, but the actual file describes an ADHD-focused software project manager. This mismatch can cause the wrong skill to be installed or trusted in a construction workflow, leading users to disclose sensitive operational data to a skill that is not designed, reviewed, or scoped for that domain.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The inline text explicitly states the skill is not for construction foremen, directly contradicting the manifest-level positioning. This is dangerous because it signals packaging or provenance confusion, making it easier for users or systems to rely on an unsuitable skill in safety-sensitive construction contexts.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · install.sh (reported line 30)May include surrounding context.

sh
cp "$SRC_DIR/SKILL.md" "$SRC_DIR/config.yaml" "$SRC_DIR/README.md" "$SRC_DIR/install.sh" "$SRC_DIR/.env.example" "$TARGET/"
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true
echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · install.sh (reported line 31)May include surrounding context.

sh
cp "$SRC_DIR/SKILL.md" "$SRC_DIR/config.yaml" "$SRC_DIR/README.md" "$SRC_DIR/install.sh" "$SRC_DIR/.env.example" "$TARGET/"
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true
echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · install.sh (reported line 32)May include surrounding context.

sh
cp "$SRC_DIR/SKILL.md" "$SRC_DIR/config.yaml" "$SRC_DIR/README.md" "$SRC_DIR/install.sh" "$SRC_DIR/.env.example" "$TARGET/"
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true
echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · install.sh (reported line 31)May include surrounding context.

sh
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true
echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля
echo ""

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The onboarding text tells the user to write to the bot in Russian using a fixed Russian prompt format, and the rest of the README is entirely localized to Russian. There is no indication that other languages are supported, optional, or intentionally restricted for a documented regional-compliance reason.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The README explicitly promotes handling photo documentation, reports, object details, crew data, and executive documentation, but it provides no warning about sensitive project, personal, or potentially confidential construction data. This can lead users to input employee names, contact details, site photos, access-relevant information, and internal project records into the skill without understanding retention, sharing, or redaction risks.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill description explicitly states 'Русский язык нативно' and says it uses Russian triggers, then the role instructions continue entirely in Russian without offering any language choice. This creates a language/locale policy issue because the skill appears to require a specific language rather than allowing user preference or documenting a justified regional restriction.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
79% confidence
Finding

The script performs rm -rf "$(dirname "$STAGE_DIR")", which recursively deletes the parent directory of the staging path. Although this is intended cleanup, there is no confirmation prompt and no explicit warning comment or user-facing notice near the deletion to disclose that a recursive removal will occur.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The markdown lists activation phrases such as "Status report", "Risk check", "Sprint plan", and "Retro" without narrowing context or defining when they should or should not invoke the skill. These are common workplace phrases and, without exclusions or explicit invocation constraints, could cause unintended activation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The line "Нет русского языка" indicates a hard language limitation in natural-language policy terms. The file does not offer a language choice, opt-in, or a documented regional/compliance reason for the restriction.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

This markdown repeatedly contrasts the skill as supporting 'Русский язык' while competitors do not, and later states competitors are 'EN-only'. That frames a language constraint as part of the skill behavior, but the file provides no indication that users can opt into another language or that the Russian-only scope is explicitly justified as a region-specific tool.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

Line L40 says 'Русский язык — все verified конкуренты EN-only', which indicates a fixed language posture as a core differentiator. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale constraint is clearly documented and justified.

Content

No source excerpt is available for this finding.