Back to skill

Security audit

plan-it

Security checks for vulnerabilities and agentic risk

Overview

This planning skill is broadly coherent, but it needs review because it reads local agent session stores and has unsafe HTML/path handling that could affect files or browser-opened plans.

Review before installing. Use it only for trusted projects or trusted plan data, consider setting PLANIT_NO_HISTORY=1 to disable cross-agent history inspection, avoid opening plans built from untrusted content, and do not pass untrusted template or slug values to the init scripts.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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 (4)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
scripts/session-catchup.py:80
Finding

Cross-Agent Session Store Access Exceeds Least Privilege

Content
View full analysis
list[str]: out: list[str] = [] xdg = os.environ.get("XDG_DATA_HOME") or str(Path.home() / ".local" / "share") db_path = Path(xdg) / "opencode" / "opencode.db" if not db_path.is_file(): return out try: with sqlite3.connect(f"file:{db_path}?mode=ro", uri=True) as conn: cur = conn.cursor() cur.execute("PRAGMA table_info(session)") cols = [r[1] for r in cur.fetchall()] if "directory" not in cols: return out cur.execute( "SELECT id, time_created FROM session WHERE directory = ? ORDER BY time_created DESC LIMIT 1", (str(cwd),), ) row = cur.fetchone() if row: age_min = (time.time() - row[1]) / 60 out.append(f"[catchup] OpenCode session {row[0]}, last touched {age_min:.0f} min ago") except sqlite3.Error: pass return out CODEX_READ_CAP = 256 * 1024 def _codex_catchup(cwd: Path) -> list[str]: out: list[str] = [] codex_dir = Path.home() / ".codex" / "sessions" if not codex_dir.is_dir(): return out files = sorted(codex_dir.glob("*.json"), key=lambda p: p.stat().st_mtime, reverse=True) cwd_str = str(cwd) for f in files[:10]: try: with f.open("rb") as fh: blob = fh.read(CODEX_READ_CAP) data = json.loads(blob.decode("utf-8", errors="replace")) except (OSError, json.JSONDecodeError, UnicodeDecodeError): continue if isinstance(data, dict) and data.get("cwd") == cwd_str: age_min = (time.time() - f.stat().st_mtime) / 60 out.app ...[truncated 2311 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
templates/implementation-plan.html:681
Finding

Stored DOM XSS Through Unsanitized Plan Data

Content
View full analysis
{ const wrap = document.createElement("div"); wrap.className = "mock"; const title = document.createElement("div"); title.className = "mock-title"; title.textContent = m.title || "Mockup"; wrap.appendChild(title); const body = document.createElement("div"); // Mockup HTML is plan-author-provided. Treat as trusted: it ships inline in plan.html. body.innerHTML = m.html || ""; wrap.appendChild(body); mocksRoot.appendChild(wrap); }); ``` `templates/living-design-system.html` does the same for component variants: ```javascript (c.html_variants || []).forEach((v) => { const vbox = document.createElement("div"); vbox.className = "variant"; const lab = document.createElement("span"); lab.className = "v-label"; lab.textContent = v.variant || "default"; const stage = document.createElement("div"); stage.className = "v-stage"; // Component HTML is plan-author-provided. Treat as trusted: ships inline in plan.html. stage.innerHTML = v.html || ""; vbox.appendChild(lab); vbox.appendChild(stage); variants.appendChild(vbox); }); ``` `templates/module-map.html` inserts multiple ordinary JSON fields into HTML without escaping: ```javascript const nodesTable = document.getElementById("nodes-table"); (plan.nodes || []).forEach(n => { const tr = document.createElement("tr"); tr.innerHTML = `${n.id}${n.label || ""}${n.kind || ""}${n.hot ? "yes" : ""}${n.x}, ${n.y}`; nodesT ...[truncated 3266 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
templates/annotated-pr.html:491
Finding

Unvalidated URL Schemes Permit Script URLs and Malicious Navigation

Content
View full analysis
${escapeHtml(pr.title)}`); if (pr.base && pr.head) parts.push(`${escapeHtml(pr.base)} <- ${escapeHtml(pr.head)}`); if (pr.author) parts.push(`@${escapeHtml(pr.author)}`); if (pr.url) parts.push(`${escapeHtml(pr.url)}`); summary.innerHTML = parts.join(" · "); ``` `templates/weekly-status.html` directly assigns a plan-controlled link: ```javascript const a = document.createElement("a"); a.href = it.link; a.target = "_blank"; a.rel = "noreferrer noopener"; ``` ### Technical Analysis Escaping attribute delimiters prevents an attacker from breaking out of the `href` attribute, but it does not make the URL safe. Browsers interpret schemes such as `javascript:` and may execute the embedded JavaScript when the link is activated. Likewise, direct assignment to `a.href` normalizes a URL but does not reject dangerous schemes. The `noopener` and `noreferrer` relationship attributes reduce opener-based attacks and referrer disclosure for new windows, but they do not validate the destination or prevent malicious navigation. A malicious plan can therefore present a legitimate-looking pull request or status link whose actual destination is an active script URL or phishing page. ### Attack Path 1. An attacker controls `plan.pr.url` or a weekly status item's `link`. 2. The plan renderer creates an anchor without scheme or origin validation. 3. The generated page displays the link as part of an apparently trusted planning ...[truncated 563 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/init-plan.sh:36
Finding

Path Traversal in Template and Parallel-Plan Names

Content
View full analysis
&2; exit 2 ;; esac done if [ -z "$TEMPLATE" ]; then echo "Available templates:" ls -1 "$RESOLVED_TEMPLATES" | sed 's/\.html$//' | sed 's/^/ /' echo read -r -p "Choose template: " TEMPLATE fi SRC="${RESOLVED_TEMPLATES}/${TEMPLATE}.html" if [ ! -f "$SRC" ]; then echo "[plan-it] template not found: $TEMPLATE" >&2 exit 3 fi if [ -n "$SLUG" ]; then DEST_DIR=".planning/${SLUG}" mkdir -p "$DEST_DIR" DEST="${DEST_DIR}/plan.html" mkdir -p ".planning" echo "$SLUG" > ".planning/.active_plan" else DEST="plan.html" fi if [ -f "$DEST" ]; then TS=$(date +%Y%m%d-%H%M%S) cp "$DEST" "${DEST}.backup.${TS}" fi cp "$SRC" "$DEST" ``` The PowerShell implementation has the same issue: ```powershell $Src = Join-Path $ResolvedTemplates "$Template.html" if (-not (Test-Path $Src)) { Write-Error "[plan-it] template not found: $Template" exit 3 } if ($Slug) { $DestDir = Join-Path ".planning" $Slug New-Item -ItemType Directory -Force -Path $DestDir | Out-Null $Dest = Join-Path $DestDir "plan.html" New-Item -ItemType Directory -Force -Path ".planning" | Out-Null Set-Content -Path ".planning\.active_plan" -Value $Slug -NoNewline } else { $Dest = "plan.html" } if (Test-Path $Dest) { $Ts = Get-Date -Format "yyyyMMdd-HHmmss" Copy-Item -Path $Dest -Destination "$Dest.backup.$Ts" } Copy-Item -Path $Src -Destination $Dest -Force ``` ### Technical Analysis Neither `TEMPLATE` nor `SLUG` is restricted to a safe identifier. Both are concatenated or joined i ...[truncated 1975 chars]
Remediation
View remediation
Vulnerability Patterns
  • 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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (36)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a full planning skill whose main function is to generate and manage a self-contained interactive HTML planning artifact. The supplied code chunk does not do that. Its primary purpose is attestation: resolve an existing plan.html path, compute a SHA-256 hash, optionally run a Python helper to seal content, and store/show/clear the resulting hash. While the description briefly mentions tamper protection by SHA-256, this script only covers that narrow integrity function and none of the broader planning, rendering, templating, export, or cross-IDE behaviors. Therefore the code chunk materially differs from the declared purpose and should be flagged as a mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description is for a user-facing planning skill centered on producing and maintaining an interactive HTML planning document. The supplied code instead is a maintenance script for updating version numbers across repository files. Its operations are limited to parsing a semver argument, locating a predefined set of files, regex-replacing version fields, validating JSON after replacement, and optionally writing changes. There is no HTML generation, no planning workflow logic, no persistence/session recovery, no cryptographic hashing, no template system, and no Markdown export. While the file list includes multiple IDE adapter paths, that is only part of version synchronization, not the described planning functionality. Therefore the code's actual behavior is materially different from the declared purpose.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a broad HTML-first planning skill centered on creating and maintaining an interactive persistent plan.html with many features. The supplied code chunk implements only one narrow supporting feature mentioned in that description: exporting embedded plan data from plan.html back into Markdown. It does not create plans, render interactive UI, protect integrity, mirror across IDEs, or handle persistence/catchup. Since the actual code’s primary purpose is materially narrower and different from the declared overall skill behavior, this chunk does not accurately represent the declared description on its own.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The code chunk is a simple scaffolding initializer: it resolves a template directory, prompts for or accepts a template name, copies an existing .html template into place, optionally under .planning//plan.html, and backs up any preexisting file. This supports only a narrow subset of the description: creating plan.html from a template, and partially aligns with the claim of shipping templates. However, nearly all prominent declared capabilities are absent from this code: there is no logic for interactive plan construction, embedded state handling, persistence across clears, integrity hashing, IDE mirroring, or Markdown export. The declared purpose therefore materially overstates and misrepresents what this supplied code actually does.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a full planning skill centered on creating and maintaining an interactive self-contained HTML planning artifact with many UX features and ecosystem behaviors. The supplied code chunk is only a backend/helper hook script for reading an existing plan.html, extracting its embedded JSON, sanitizing and summarizing it for agent hooks, checking completion/evidence policy, and handling attestation/sealing verification. Some overlap exists around tamper/integrity concepts and embedded JSON state, but the primary purpose is materially different: this script does not create the HTML plan, render UI, provide templates/categories, mirror across IDEs, survive /clear, or export to Markdown. It also has undeclared operational capabilities such as stop-hook gating and injection of summarized plan data into the agent context. Therefore the description does not accurately represent what this code chunk actually does.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description presents a full planning system whose core behavior is generating and managing a sophisticated persistent HTML planning artifact. The supplied code chunk does something much narrower: it resolves the path to an already-existing plan.html and opens it in the default browser. While opening a rendered plan could be a supporting utility within such a skill, this specific code does not implement the vast majority of the described functionality. Its primary purpose is materially different from the declared purpose of generating and maintaining the planning artifact. It also invokes external OS/browser-opening commands, which is an actual capability not reflected in the description. Therefore this chunk does not accurately represent the declared behavior.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a broad planning skill centered on creating and managing an interactive HTML planning artifact. The supplied code chunk does not perform that core function. Instead, it is a narrow support script whose sole behavior is to detect possible prior sessions for the same project by reading local metadata from Claude Code, OpenCode, and Codex session stores and then printing a catchup notice. While 'survives /clear via session catchup' is one feature mentioned in the description, this code only implements that small piece and none of the primary advertised planning capabilities. It also accesses local session storage and a SQLite database across tools, which is a materially different resource access pattern than implied by the empty declared permissions. Therefore the code chunk is a significant mismatch with the declared purpose.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description is for a user-facing planning skill centered on creating and managing a persistent interactive HTML plan. The supplied code chunk instead is a maintenance script whose sole function is to copy or verify canonical scripts and HTML templates across a set of IDE-specific skill directories. While the description mentions mirroring across IDEs, that is only one supporting aspect of the broader declared skill. This code does not implement the core planning behavior, artifact generation, persistence, interactivity, session recovery, tamper protection, or Markdown export. Its primary purpose is materially different from the declared purpose, so this is a clear mismatch.

Content

No source excerpt is available for this finding.

Lp1

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'env' capability that is not listed in its permissions. This may indicate deceptive intent or missing permission declarations.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code injects plan-author-provided HTML directly with innerHTML while explicitly treating it as trusted, which creates a straightforward DOM XSS sink if the embedded JSON is modified or user-controlled. The skill context makes this more dangerous because the artifact is meant to persist and be mirrored/exported, so malicious markup could survive across sessions and environments and execute wherever the HTML plan is opened.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest says the plan is 'tamper-protected,' but later documentation correctly downgrades this to tamper-evident and explicitly admits a modified plan can be re-sealed. This can mislead users into treating the integrity mechanism as a security control that prevents tampering, when it only signals divergence and can be deliberately refreshed by the same actor modifying content.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation guidance is very broad: it triggers not only on explicit phrases like 'plan it' but also on 'starting any multi-step task.' In a system with powerful hooks and tool permissions, an overbroad trigger can cause the skill to activate unexpectedly, inject context, influence workflow decisions, and modify planning artifacts in scenarios where the user did not ask for it.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
88% confidence
Finding

The hook uses shell path probing with ls over user-local skill/plugin directories to locate helper scripts. Even though the paths are narrow, this performs local environment enumeration and reveals installation layout; in adversarial skill ecosystems, such probing can leak metadata about installed tooling and expand the effective knowledge surface without explicit user intent.

Content

Scanner excerpt · SKILL.md (reported line 10)May include surrounding context.

md
UserPromptSubmit:
    - hooks:
        - type: command
          command: "if [ -f plan.html ]; then PY=$(command -v python3 || command -v python); HELPER=\"${CLAUDE_PLUGIN_ROOT:-$HOME/.claude/skills/plan-it}/scripts/plan-hook.py\"; if [ -z \"$PY\" ] || [ ! -f \"$HELPER\" ]; then HELPER=$(ls \"$HOME/.claude/skills/plan-it/scripts/plan-hook.py\" \"$HOME/.claude/plugins/marketplaces/plan-it/scripts/plan-hook.py\" 2>/dev/null | head -1); fi; if [ -n \"$PY\" ] && [ -n \"$HELPER\" ] && [ -f \"$HELPER\" ]; then \"$PY\" \"$HELPER\" inject --mode summary --lines 30; else echo '[plan-it] plan.html present but helper script not found. Run /plan to re-init.'; fi; fi"
  PreToolUse:
    - matcher: "Write|Edit|Bash|Read|Glob|Grep"
      hooks:

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
88% confidence
Finding

This repeated helper lookup in PreToolUse runs frequently and therefore increases the amount of local environment probing during normal operation. Repeated enumeration across tool events raises privacy and predictability concerns and makes the skill's behavior more invasive than a simple planning helper.

Content

Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.

md
- matcher: "Write|Edit|Bash|Read|Glob|Grep"
      hooks:
        - type: command
          command: "if [ -f plan.html ]; then PY=$(command -v python3 || command -v python); HELPER=\"${CLAUDE_PLUGIN_ROOT:-$HOME/.claude/skills/plan-it}/scripts/plan-hook.py\"; if [ -z \"$PY\" ] || [ ! -f \"$HELPER\" ]; then HELPER=$(ls \"$HOME/.claude/skills/plan-it/scripts/plan-hook.py\" \"$HOME/.claude/plugins/marketplaces/plan-it/scripts/plan-hook.py\" 2>/dev/null | head -1); fi; if [ -n \"$PY\" ] && [ -n \"$HELPER\" ] && [ -f \"$HELPER\" ]; then \"$PY\" \"$HELPER\" inject --mode active-phase --lines 15; fi; fi"
  PostToolUse:
    - matcher: "Write|Edit"
      hooks:

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
88% confidence
Finding

The Stop hook repeats the same directory enumeration pattern, again disclosing local installation structure to the skill runtime. While not critical on its own, repeated environment discovery across lifecycle hooks is unnecessary and broadens local reconnaissance capability.

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

md
Stop:
    - hooks:
        - type: command
          command: "if [ -f plan.html ]; then PY=$(command -v python3 || command -v python); HELPER=\"${CLAUDE_PLUGIN_ROOT:-$HOME/.claude/skills/plan-it}/scripts/plan-hook.py\"; if [ -z \"$PY\" ] || [ ! -f \"$HELPER\" ]; then HELPER=$(ls \"$HOME/.claude/skills/plan-it/scripts/plan-hook.py\" \"$HOME/.claude/plugins/marketplaces/plan-it/scripts/plan-hook.py\" 2>/dev/null | head -1); fi; if [ -n \"$PY\" ] && [ -n \"$HELPER\" ] && [ -f \"$HELPER\" ]; then \"$PY\" \"$HELPER\" check-complete; fi; fi"
  PreCompact:
    - matcher: "*"
      hooks:

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
88% confidence
Finding

The PreCompact hook performs the same helper-path enumeration and does so during context compaction, a sensitive lifecycle stage where minimizing extra behavior is preferable. This compounds the environment-discovery pattern and may surface local installation details across multiple events.

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

md
- matcher: "*"
      hooks:
        - type: command
          command: "if [ -f plan.html ]; then echo '[plan-it] PreCompact: context compaction about to occur.'; echo 'Before compaction completes: ensure plan.html embedded JSON captures recent progress_log entries and current_phase status.'; echo 'plan.html remains on disk and will be re-read after compaction.'; PY=$(command -v python3 || command -v python); HELPER=\"${CLAUDE_PLUGIN_ROOT:-$HOME/.claude/skills/plan-it}/scripts/plan-hook.py\"; if [ -z \"$PY\" ] || [ ! -f \"$HELPER\" ]; then HELPER=$(ls \"$HOME/.claude/skills/plan-it/scripts/plan-hook.py\" \"$HOME/.claude/plugins/marketplaces/plan-it/scripts/plan-hook.py\" 2>/dev/null | head -1); fi; if [ -n \"$PY\" ] && [ -n \"$HELPER\" ] && [ -f \"$HELPER\" ]; then \"$PY\" \"$HELPER\" attestation; fi; fi; exit 0"
metadata:
  version: "0.2.0"
---

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documentation itself warns not to overclaim the seal, yet the manifest still uses stronger language that implies protection rather than evidence. Contradictory security messaging increases the chance that users or downstream agents will trust forged or re-sealed plan state more than they should.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 26)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 27)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 28)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 29)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 30)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 31)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 32)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · scripts/bump-version.py (reported line 33)May include surrounding context.

python
# were not actually translated. They return in v0.2.0 with real translations.
PARITY_FILES = [
    # 12 SKILL.md files (1 canonical + 11 IDE adapters)
    "skills/plan-it/SKILL.md",
    ".codex/skills/plan-it/SKILL.md",
    ".cursor/skills/plan-it/SKILL.md",
    ".codebuddy/skills/plan-it/SKILL.md",