Back to skill

Security audit

chronos

Security checks for vulnerabilities and agentic risk

Overview

Chronos has a coherent time-tracking purpose, but it installs broad persistent agent hooks and has unsafe installer/path handling that users should review before installing.

Install only if you want a persistent local activity ledger and automatic agent hooks. Prefer project-scoped installation where possible, inspect the exact hook commands before enabling them, avoid installing from paths with spaces or shell metacharacters, and pin/review any package-based installer source.

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 (4)

T09 · Insecure Skill Coding Practices

Error
Location
installers/detect-shell.sh:3
Finding

Shell Command Injection Through Unescaped eval Output

Content
View full analysis
set -euo pipefail ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)" if command -v bash >/dev/null 2>&1 && [ "${OS:-}" != "Windows_NT" ] || [ -n "${WSL_DISTRO_NAME:-}" ]; then echo "export CHRONOS_SHELL=bash" echo "export CHRONOS_SETTINGS_FRAGMENT=$ROOT/installers/claude-code/settings.json" elif command -v pwsh >/dev/null 2>&1 || command -v powershell >/dev/null 2>&1; then echo "export CHRONOS_SHELL=powershell" echo "export CHRONOS_SETTINGS_FRAGMENT=$ROOT/installers/claude-code/settings.windows-ps.json" else echo "export CHRONOS_SHELL=bash" echo "export CHRONOS_SETTINGS_FRAGMENT=$ROOT/installers/claude-code/settings.json" fi ``` ### Technical Analysis The script explicitly recommends evaluating its output: ```bash eval "$(./installers/detect-shell.sh)" ``` It includes the repository path in generated shell assignments without shell escaping. Although `ROOT` is quoted while being calculated, its value is later embedded verbatim in a string that is intended to be parsed again by the shell. If the repository is located under a directory whose name contains shell metacharacters, command substitution, a semicolon, whitespace, or a newline, those characters can become active shell syntax during `eval`. Quoting at the time `ROOT` is calculated does not protect the value during this second parsing stage. ### Attack Path 1. An attacker supplies or recommends a repository archive or checkout located under a crafted directory name containing shell syntax. 2. The user follows the documented usage: ```bash ...[truncated 902 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
installers/claude-code/install.sh:96
Finding

Persistent Hook Command Injection Through Unquoted Installation Path

Content
View full analysis
"$TMP" ``` The substituted values are used as unquoted shell commands: ```json { "hooks": { "SessionStart": [ { "matcher": "", "hooks": [ { "type": "command", "command": "bash ${CHRONOS_ROOT}/scripts/session_start.sh" } ] } ], "UserPromptSubmit": [ { "matcher": "", "hooks": [ { "type": "command", "command": "bash ${CHRONOS_ROOT}/scripts/prompt_submit.sh" } ] } ], "PreToolUse": [ { "matcher": "", "hooks": [ { "type": "command", "command": "bash ${CHRONOS_ROOT}/scripts/pre_tool.sh", "async": true } ] } ], "PostToolUse": [ { "matcher": "", "hooks": [ { "type": "c ...[truncated 3089 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/_lib.sh:49
Finding

Unvalidated Session Identifiers Used in Ledger and State File Paths

Content
View full analysis
/dev/null || true) if [ -z "${sid:-}" ]; then local file="$CHRONOS_HOME/current-session" if [ -f "$file" ]; then sid=$(cat "$file") else sid="anon-$(now_epoch)" echo "$sid" > "$file" fi fi echo "$sid" } ``` Session startup then truncates the derived ledger path and writes the state path: ```bash INPUT=$(cat) SESSION=$(resolve_session_id "$INPUT") LEDGER=$(ledger_path "$SESSION") STATE=$(state_path "$SESSION") NOW_UTC=$(now_utc_iso) NOW_LOCAL=$(now_local_iso) NOW_EPOCH=$(now_epoch) TZ=$(tz_name) OFFSET=$(utc_offset) # Init ledger (truncate if fresh session start) : > "$LEDGER" ``` The PowerShell implementation follows the same pattern: ```powershell function Get-LedgerPath { param($Session) Join-Path $script:ChronosHome "ledger-$Session.jsonl" } function Get-StatePath { param($Session) Join-Path $script:ChronosHome "session-$Session.json" } function Resolve-SessionId { param([string]$Input) $sid = Read-JsonField -Field 'session_id' -Input $Input if (-not $sid) { $file = Join-Path $script:ChronosHome 'current-session' if (Test-Path $file) { $sid = Get-Content $file -Raw } else { $sid = ...[truncated 3399 chars]
Remediation
View remediation
&2 return 1 fi ``` 2. Apply equivalent checks in PowerShell and TypeScript: ```typescript if (!/^[A-Za-z0-9_-]{1,128}$/.test(sid)) { throw new Error("Invalid session ID"); } ``` 3. Prefer deriving the on-disk filename from a cryptographic hash of the external session ID: ```typescript const safeSid = crypto.createHash("sha256").update(sid).digest("hex"); ``` 4. Resolve the final canonical path and verify containment beneath the canonical `CHRONOS_HOME` directory before every write. 5. Reject: - `/` and `\` - `..` - Absolute paths - NUL and control characters - Excessively long identifiers 6. Open files with restrictive permissions and avoid following symbolic links where the platform supports secure no-follow options. 7. Add tests for Unix traversal, Windows traversal, absolute paths, mixed separators, symbolic-link redirection, and oversized identifiers. ]]>

T08 · Insecure Dependencies

Warning
Location
README.md:12
Finding

Unpinned Third-Party Package Execution in Installation Instructions

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (46)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill description emphasizes benign temporal reasoning, but the content also describes persistent tool-use telemetry collection including tool names, timestamps, success state, and hashed arguments in a session ledger. That creates a real privacy and security concern because command metadata and argument fingerprints can reveal sensitive operational activity, and users may not realize such telemetry is being retained from the headline description alone.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill description emphasizes benign temporal reasoning, but the content also describes persistent tool-use telemetry collection including tool names, timestamps, success state, and hashed arguments in a session ledger. That creates a real privacy and security concern because command metadata and argument fingerprints can reveal sensitive operational activity, and users may not realize such telemetry is being retained from the headline description alone.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The skill description emphasizes benign temporal reasoning, but the content also describes persistent tool-use telemetry collection including tool names, timestamps, success state, and hashed arguments in a session ledger. That creates a real privacy and security concern because command metadata and argument fingerprints can reveal sensitive operational activity, and users may not realize such telemetry is being retained from the headline description alone.

Content

No source excerpt is available for this finding.

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
95% confidence
Finding

This installer explicitly modifies Claude Code's trusted configuration file under ~/.claude/settings.json to register hooks that will execute later in the agent environment. Writing to an agent config directory is security-sensitive because it creates persistence and can alter future agent behavior; in this skill's context, that is the core installation mechanism, so the finding is real even if not obviously malicious.

Content

Scanner excerpt · installers/claude-code/install.sh (reported line 3)May include surrounding context.

sh
#!/usr/bin/env bash
# chronos — Claude Code installer
# Concatenates chronos hooks into ~/.claude/settings.json (or .claude/settings.json with --project).
# Uses array concat per hook event so existing hooks (mempalace, planning-with-files, etc.) are preserved.

set -euo pipefail

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
97% confidence
Finding

The script creates and writes to either project-local .claude/settings.json or user-global $HOME/.claude/settings.json, which is a privileged agent control surface. Because this skill installs hook-backed behavior for autonomous or long-running agents, modifying that file can influence subsequent executions and persistence, making the access materially more dangerous than ordinary application config writes.

Content

Scanner excerpt · installers/claude-code/install.sh (reported line 40)May include surrounding context.

sh
if [ "$SCOPE" = project ]; then
    mkdir -p .claude; TARGET=.claude/settings.json
  else
    mkdir -p "$HOME/.claude"; TARGET="$HOME/.claude/settings.json"
  fi
fi

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · CONTRIBUTING.md (reported line 26)May include surrounding context.

md
{
  "_comment": "chronos — drop these entries into ~/.claude/settings.json or .claude/settings.json. Replace ${CHRONOS_ROOT} with the absolute path to your chronos checkout, e.g. /c/Users/you/Documents/here-now-projects/chronos. On Windows with Git Bash, .sh paths work. For pure PowerShell, use the .ps1 variants below and set shell to powershell.",
  "hooks": {
    "SessionStart": [
      {

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · README.md (reported line 124)May include surrounding context.

md
{
  "_comment": "chronos — drop these entries into ~/.claude/settings.json or .claude/settings.json. Replace ${CHRONOS_ROOT} with the absolute path to your chronos checkout, e.g. /c/Users/you/Documents/here-now-projects/chronos. On Windows with Git Bash, .sh paths work. For pure PowerShell, use the .ps1 variants below and set shell to powershell.",
  "hooks": {
    "SessionStart": [
      {

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · installers/claude-code/settings.json (reported line 2)May include surrounding context.

json
{
  "_comment": "chronos — drop these entries into ~/.claude/settings.json or .claude/settings.json. Replace ${CHRONOS_ROOT} with the absolute path to your chronos checkout, e.g. /c/Users/you/Documents/here-now-projects/chronos. On Windows with Git Bash, .sh paths work. For pure PowerShell, use the .ps1 variants below and set shell to powershell.",
  "hooks": {
    "SessionStart": [
      {

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest configures automatic execution of PowerShell scripts with -ExecutionPolicy Bypass on multiple lifecycle events, including prompt and tool activity. This suppresses a core Windows safety control and creates a strong code-execution primitive: if the referenced scripts or installation path are modified, malicious code will run automatically and repeatedly with the user's privileges.

Content

No source excerpt is available for this finding.

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · README.md (reported line 138)May include surrounding context.

md
{
  "_comment": "chronos — Codex CLI hooks. Drop into ~/.codex/hooks.json (user) or <repo>/.codex/hooks.json. Also add [features] codex_hooks = true to ~/.codex/config.toml. Replace ${CHRONOS_ROOT}. Note: Codex PreToolUse matches Bash only; Read/Write/Edit tool events may be unavailable until coverage extends.",
  "hooks": {
    "SessionStart": [
      { "command": "bash ${CHRONOS_ROOT}/scripts/session_start.sh" }

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · installers/codex/config.toml.fragment (reported line 2)May include surrounding context.

text
{
  "_comment": "chronos — Codex CLI hooks. Drop into ~/.codex/hooks.json (user) or <repo>/.codex/hooks.json. Also add [features] codex_hooks = true to ~/.codex/config.toml. Replace ${CHRONOS_ROOT}. Note: Codex PreToolUse matches Bash only; Read/Write/Edit tool events may be unavailable until coverage extends.",
  "hooks": {
    "SessionStart": [
      { "command": "bash ${CHRONOS_ROOT}/scripts/session_start.sh" }

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · installers/codex/hooks.json (reported line 2)May include surrounding context.

json
{
  "_comment": "chronos — Codex CLI hooks. Drop into ~/.codex/hooks.json (user) or <repo>/.codex/hooks.json. Also add [features] codex_hooks = true to ~/.codex/config.toml. Replace ${CHRONOS_ROOT}. Note: Codex PreToolUse matches Bash only; Read/Write/Edit tool events may be unavailable until coverage extends.",
  "hooks": {
    "SessionStart": [
      { "command": "bash ${CHRONOS_ROOT}/scripts/session_start.sh" }

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · installers/codex/install.sh (reported line 39)May include surrounding context.

sh
mv "$TMP" "$TARGET"

# Enable feature flag
CONF="$HOME/.codex/config.toml"
[ -f "$CONF" ] || touch "$CONF"
if ! grep -q '^\s*codex_hooks\s*=\s*true' "$CONF" 2>/dev/null; then
  if ! grep -q '^\[features\]' "$CONF" 2>/dev/null; then

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.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · CONTRIBUTING.md (reported line 7)May include surrounding context.

md
To add chronos support for a new agent platform:

1. Create `installers/<platform>/README.md` describing the platform's hook or plugin API
2. Write the context injection: SessionStart must emit `now_utc`, `now_local`, `tz`, `ledger`, `state`, `session`
3. Write per-turn injection: UserPromptSubmit must emit `turn`, `session_duration`, `since_last_user`
4. Write the ledger writer: PreToolUse appends `started_at`, PostToolUse appends `finished_at + duration_ms + success`

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · README.md (reported line 23)May include surrounding context.

md
## The problem

LLMs do not know what time it is. They guess. They report "just now" when it was two hours ago. They retry failed commands without checking that the last attempt was 10 seconds ago. They run silently in autonomous loops with no idle detection.

[Arxiv 2510.23853](https://arxiv.org/abs/2510.23853) measured this empirically: even frontier models with timestamps injected into context reach only **65% temporal reasoning alignment** with human judgment. Timestamps appear in fewer than 4% of reasoning traces. The ceiling exists because having access to timestamps is not enough. Agents need explicit rules for *when to look at the clock*.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 143)May include surrounding context.

OpenCode

bash
mkdir -p ~/.config/opencode/plugins
cp installers/opencode/plugin.ts ~/.config/opencode/plugins/chronos.ts
mkdir -p ~/.config/opencode/skills/chronos
cp SKILL.md ~/.config/opencode/skills/chronos/SKILL.md

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest description says the skill activates whenever an agent reasons about recency, retry windows, memory staleness, deploy cooldowns, idle detection, or any 'when / how long ago' question. That scope is very broad and overlaps with ordinary, common time-related phrasing, without clear exclusion conditions or tighter invocation constraints.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · installers/adal/README.md (reported line 9)May include surrounding context.

bash
# path may vary by ADAL version — check docs
mkdir -p ~/.adal/skills/chronos
cp SKILL.md ~/.adal/skills/chronos/SKILL.md
cp AGENTS.md ~/.adal/skills/chronos/AGENTS.md

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

All hook matchers are empty strings, which effectively scopes these commands to every occurrence of the corresponding Claude Code lifecycle event. That causes the referenced scripts to run broadly across sessions, prompts, and tool activity, increasing blast radius if the scripts are buggy, modified, or malicious and making behavior hard to reason about.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

All hook entries use an empty matcher, so the configured commands will run for every matching lifecycle event without any scoping. In a skill designed to add temporal awareness, broad triggering may be intentional, but it still increases attack surface, causes unnecessary execution frequency, and makes any downstream script bug or compromise fire on all sessions, prompts, and tool events.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This manifest installs automatic shell command execution on routine agent events such as session start, prompt submission, and tool use, but the embedded guidance does not provide a clear user-facing warning about the frequency and implications of those executions. That creates a meaningful trust and transparency problem: users may enable the hooks without understanding that arbitrary local shell scripts will run repeatedly during normal operation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

This manifest wires commands to generic lifecycle events such as SessionStart, UserPromptSubmit, PreToolUse, PostToolUse, and Stop without describing any limiting conditions for when the hooks should or should not run. Because manifest files are in scope for vague-trigger review, the absence of narrower trigger constraints or exclusions can lead to unintended invocation across ordinary CLI activity.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · installers/cursor/README.md (reported line 8)May include surrounding context.

Install (project)

bash
mkdir -p .cursor/rules
cp .cursor/rules/chronos.mdc .cursor/rules/chronos.mdc