Back to skill

Security audit

Succession Plan

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed local succession-task tracker with no network behavior, but users should treat its plain-text local logs as sensitive.

Install only if a local plain-text task log is acceptable for your succession planning notes. Avoid entering confidential personnel assessments on shared machines, review the data directory periodically, and be aware that one auxiliary script has a low-risk wildcard handling bug that may print local filenames if given wildcard input.

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

Note
Location
scripts/succession.sh:28
Finding

Unquoted Argument Expansion Allows Local Filename Disclosure

Content
View full analysis
/dev/null || true; INPUT="$*" python3 -c ' import sys cmd=sys.argv[1] if len(sys.argv)>1 else "help" inp=" ".join(sys.argv[2:]) if cmd=="plan": role=inp if inp else "Key Role" print("=" * 55) print(" Succession Plan: {}".format(role)) print("=" * 55) for section in [("Current Holder","Name: ___ Tenure: ___ Retirement/departure: ___"),("Critical Knowledge","1. ___\n 2. ___\n 3. ___"),("Successor Candidates","Primary: ___ (Readiness: Now/6mo/1yr)\n Secondary: ___ (Readiness: ___)\n External: Consider if needed"),("Development Plan","1. Mentoring with current holder\n 2. Stretch assignments\n 3. Training courses\n 4. Cross-functional exposure"),("Transition Timeline","Month 1: Shadow and observe\n Month 2-3: Co-lead responsibilities\n Month 4-6: Lead with support\n Month 7+: Fully independent")]: print("\n {}:".format(section[0])) for line in section[1].split("\n"): print(" {}".format(line.strip())) elif cmd=="matrix": print(" 9-Box Talent Matrix:") print(" Performance ->") print(" Low Med High") print(" High | Enigma | Growth Star | Future Leader |") print(" Med | Underperf | Core Player | High Potential|") print(" Low | Risk | Average | Specialist |") print(" ^ Potential") elif cmd=="help": print("Succession Planner\n plan [role] — Succession plan template\n matrix — 9-Box talent matrix") else: print("Unknown: "+cmd) print("\nPowered by BytesAgain | bytesagain.com") ' "$CMD" $INPUT ``` ### Technical Analysis The script collects all remaining command-line arguments in `INPUT` and later expands that variable without quotation: ```bash ' "$ ...[truncated 2178 chars]
Remediation
View remediation
/dev/null || true python3 -c ' # Existing Python implementation ' "$CMD" "$@" ``` `"$@"` expands each supplied argument as a distinct, quoted argument without shell word splitting or pathname expansion. Alternatively, if all remaining arguments must intentionally be passed as one string, quote the existing variable: ```bash ' "$CMD" "$INPUT" ``` The argument-array approach is preferable because it preserves the caller's original argument boundaries. Additional hardening steps: 1. Run ShellCheck in CI and treat warning `SC2086` as a failure. 2. Add regression tests using literal wildcard and whitespace-containing inputs: ```bash ./scripts/succession.sh plan '*' ./scripts/succession.sh plan 'Role with spaces' ``` 3. Verify that output contains the literal supplied value and does not contain unrelated filenames from the current directory. 4. Avoid flattening arguments through `INPUT="$*"` unless a single combined value is explicitly required. ]]>
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (7)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill’s manifest and name present it as a succession-planning capability, but the documented behavior is a generic local task-management CLI with logging, export, reminders, and history capture. This mismatch is security-relevant because users or orchestrators may grant or invoke the skill under false assumptions, causing unintended data collection, storage, and use outside the declared business purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The implementation materially contradicts the declared skill purpose: it is a generic personal task manager rather than a succession-planning tool. This kind of capability mismatch is dangerous because users may supply sensitive HR or organizational planning data under false expectations, while the tool performs unrelated storage and logging behaviors outside the advertised business context.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The inline description and help text explicitly market the script as a productivity/task tool, which conflicts with the declared succession-planning metadata. In a security context, contradictory identity and documentation increase the risk of deceptive deployment, misuse, and mishandling of sensitive user inputs because operators cannot reliably understand the skill's true purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest advertises role mapping and transition-roadmap planning, while the body describes a generic task manager. In an agent ecosystem, this semantic deception can bypass user expectations or policy controls tied to the declared purpose, leading to inappropriate invocation and persistence of potentially sensitive HR-related notes in local logs.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The succession-planning framing suggests a specialized HR workflow tool, but the implementation described is a generic productivity CLI. That contradiction increases the risk of misuse in sensitive contexts, because operators may enter candidate assessments, transfer plans, or other confidential personnel data into a tool whose actual function is plain-text storage and command history logging.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The script creates a local data directory and persists task content to data.log without any user-facing disclosure, consent, or retention notice. In the stated succession-planning context, stored entries could contain sensitive personnel, promotion, transfer, or readiness information, making undisclosed at-rest storage a meaningful confidentiality risk on shared or unmanaged systems.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The _log helper records command activity and arguments to a history log without clear notice to the user. Because command arguments may include names, staffing decisions, or succession notes, silent history logging can expand exposure of sensitive organizational data beyond the primary datastore and create forensic leakage users do not expect.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.