Back to skill

Security audit

ponder

Security checks for vulnerabilities and agentic risk

Overview

This review skill automatically stores and recalls audit discussions through an optional database without clearly documented project boundaries.

For private projects, set PONDER_OCA=off before running commands. Enable database memory deliberately after checking its destination, project separation, access permissions, and retention. Review and pin the external installer and engine before execution, and account for subprocess model usage. All bundled documentation was inspected, but the external implementation was unavailable for verification.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
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
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (18)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md
- `references/protocol.md` - every command, flag, JSON/JSONL shape, verdict rule, exit code, `unit-edit` / `phantom` multi-path verification, guessed-root refus

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 71)May include surrounding context.

md
carried** against an established project root (`--project` or repo marker; see `references/protocol.md`).

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

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

md
- `references/perspectives.md` - Skeptic / Builder / Dreamer / Empath prompts and fan-out patterns.

External Model or Provider Selection

High
Category
Excessive Agency
Confidence
90% confidence
Finding

Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.

Content

Scanner excerpt · SKILL.md (reported line 99)May include surrounding context.

  • In Codex: run batches sequentially or spawn background codex exec workers (--skip-git-repo-check and </dev/null prevent non-git exits and stdin hangs):
sh
codex exec -C "$PWD" --sandbox workspace-write --skip-git-repo-check \
  -o .ponder/tmp/batch/batch-3.txt \
  "Grade these 10 screens against ~/code/ponder/lenses/ui.md. Units: ... .
   Write one JSON finding object per line to .ponder/tmp/batch/findings-3.jsonl. Output nothing else." \

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
85% 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 · references/harness-notes.md (reported line 66)May include surrounding context.

sh
ls -l ~/.claude/skills/ponder ~/.codex/skills/ponder    # both should point at ~/code/ponder/skill/ponder
head -4 ~/.claude/skills/ponder/SKILL.md                 # front matter intact through the symlink
ponder doctor                                            # engine reachable, and what it can see

External Model or Provider Selection

High
Category
Excessive Agency
Confidence
90% confidence
Finding

Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.

Content

Scanner excerpt · references/harness-notes.md (reported line 99)May include surrounding context.

codex exec flags that matter for fan-out:

sh
codex exec -C "$PWD" \
  --sandbox workspace-write \
  --skip-git-repo-check \
  -o .ponder/tmp/batch-3.txt \

Keyword Baiting Trigger

Medium
Category
Trigger Abuse
Confidence
80% confidence
Finding

Keyword Baiting Trigger: description clause 'Use when the user says "ponder " ("ponder the UI", "ponder our error handling", "ponder this PR"), "run a ponder audit", "grade every unit with the po' is designed to match all or most user inputs

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 · references/harness-notes.md (reported line 45)May include surrounding context.

text

Codex additionally understands an optional `metadata:` block - `disable-model-invocation`
(`~/.codex/skills/zoom-out/SKILL.md`), `argument-hint` (`~/.codex/skills/handoff/SKILL.md`) and
`short-description` (`~/.codex/vendor_imports/skills/skills/.curated/linear/SKILL.md`). Claude Code
reads the same keys at the top level instead, not nested - compare its own copy of zoom-out, which
carries a bare `disable-model-invocation: true`. So a `metadata:` block would make the two installs

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 · references/harness-notes.md (reported line 45)May include surrounding context.

text

Codex additionally understands an optional `metadata:` block - `disable-model-invocation`
(`~/.codex/skills/zoom-out/SKILL.md`), `argument-hint` (`~/.codex/skills/handoff/SKILL.md`) and
`short-description` (`~/.codex/vendor_imports/skills/skills/.curated/linear/SKILL.md`). Claude Code
reads the same keys at the top level instead, not nested - compare its own copy of zoom-out, which
carries a bare `disable-model-invocation: true`. So a `metadata:` block would make the two installs

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 · references/harness-notes.md (reported line 59)May include surrounding context.

sh
python3 -c "import yaml,sys; d=yaml.safe_load(open(sys.argv[1]).read().split('---',2)[1]); \
  assert set(d) == {'name','description'}, d.keys(); print('ok:', d['name'], len(d['description']))" \
  ~/code/ponder/skill/ponder/SKILL.md

Checking the install

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
85% 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 · references/harness-notes.md (reported line 65)May include surrounding context.

Checking the install

sh
ls -l ~/.claude/skills/ponder ~/.codex/skills/ponder    # both should point at ~/code/ponder/skill/ponder
head -4 ~/.claude/skills/ponder/SKILL.md                 # front matter intact through the symlink
ponder doctor                                            # engine reachable, and what it can see

File System Enumeration

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code scans file system directories looking for sensitive files. This could be reconnaissance for credential theft.

Content

Scanner excerpt · references/harness-notes.md (reported line 65)May include surrounding context.

Checking the install

sh
ls -l ~/.claude/skills/ponder ~/.codex/skills/ponder    # both should point at ~/code/ponder/skill/ponder
head -4 ~/.claude/skills/ponder/SKILL.md                 # front matter intact through the symlink
ponder doctor                                            # engine reachable, and what it can see

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 · references/harness-notes.md (reported line 66)May include surrounding context.

sh
ls -l ~/.claude/skills/ponder ~/.codex/skills/ponder    # both should point at ~/code/ponder/skill/ponder
head -4 ~/.claude/skills/ponder/SKILL.md                 # front matter intact through the symlink
ponder doctor                                            # engine reachable, and what it can see

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 · references/harness-notes.md (reported line 84)May include surrounding context.

by path and carry on rather than debating it:

sh
cat ~/code/ponder/skill/ponder/SKILL.md

Subagents and fan-out

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 · references/harness-notes.md (reported line 257)May include surrounding context.

md
the wait ran out. Re-run the command. See [The session lock](#the-session-lock); do not delete a
  `.lock` file unless the process that wrote it is gone.
- **A skill edit that does not take effect.** Both harnesses read SKILL.md at load; edit the repo copy
  (`~/code/ponder/skill/ponder/SKILL.md`) and start a new turn - in Codex, a new session.
- **The OCA's cognitive-loop process on `:3333` is not running.** Nothing in ponder depends on it; the
  bridge shells out to `psql` directly. If something suggests otherwise, it is wrong.

Session Persistence

Medium
Category
Rogue Agent
Confidence
65% 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 · references/harness-notes.md (reported line 94)May include surrounding context.

md
| In-process subagents | **Agent tool.** Multiple calls in one message run in parallel. Returns text to the parent. | None. |
| Very large sweeps | **Workflow tool** - handles batching, retries and collection better than hand-rolled parallel agents. | None. |
| Subprocess agents | possible, rarely worth it | **`codex exec`** - the primary mechanism |
| Background | Bash tool `run_in_background`, plus Monitor to wait on a condition | `&` + `wait`, or `nohup … &` and poll the output file |

`codex exec` flags that matter for fan-out:

The skill includes an OCA PostgreSQL read/write integration beyond its stated audit-ledger purpose.

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

The notes describe automatically reading prior memory and writing audit-related information to an OCA PostgreSQL store when reachable, while presenting silent degradation as normal behavior. This creates an additional persistence and cross-session data boundary beyond the project ledger without an explicit consent step, potentially exposing sensitive review arguments or importing unrelated stored context; the documented localhost destination does not establish external exfiltration.

Content

No source excerpt is available for this finding.

The manifest promises zero-churn convergence before stopping, but the protocol permits convergence with nonzero churn and allows stopping without convergence.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

L702-L704 classify two consecutive passes with churn below 0.10 and novelty below 0.05 as converged, rather than requiring zero churn. L705-L710 also permit perseveration and budget stops, while L210-L215 allow explicit stop overrides once coverage is complete. These documented stopping rules contradict the manifest's claim that zero-churn convergence is enforced before stopping, allowing an audit to terminate while findings are still changing.

Content

No source excerpt is available for this finding.