Pilot Keychain

Security checks across static analysis, malware telemetry, and agentic risk

Overview

This skill is meant to exchange credentials, but its instructions handle raw secrets in files and between agents without enough scoping, verification, or support for its security claims.

Review carefully before installing. Use only with trusted Pilot Protocol agents, confirm every recipient and credential manually, prefer short-lived scoped tokens, and do not assume the advertised auto-expiry or encryption guarantees are enforced by this SKILL.md alone.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

#
ASI03: Identity and Privilege Abuse
High
What this means

A credential sent to the wrong or untrusted agent could give that agent access to the user's accounts or services.

Why it was flagged

The skill is explicitly intended to transfer credentials, which can grant access to user or service accounts. The artifacts do not define approval, recipient verification, credential scope, or revocation requirements before sharing.

Skill content
Use this skill when:
  1. You need to share API keys, tokens, or credentials securely between agents
Recommendation

Only use this with explicitly approved recipients and scoped, revocable credentials; require confirmation before sending any real secret.

#
ASI07: Insecure Inter-Agent Communication
High
What this means

Secrets may be shared over an agent-to-agent channel without enough assurance that the recipient is the intended trusted party.

Why it was flagged

The workflow sends a file containing a credential to another Pilot Protocol agent, but the instructions do not show identity verification, trust checks, or clear data-boundary controls for the recipient.

Skill content
pilotctl --json send-file "$RECIPIENT" /tmp/cred.json
Recommendation

Verify recipient identity out of band, use least-privilege temporary credentials, and avoid autonomous sends without user confirmation.

#
ASI09: Human-Agent Trust Exploitation
Medium
What this means

Users may believe credentials are automatically expired or rotated when the artifact only demonstrates metadata and manual cleanup.

Why it was flagged

The description promises automatic expiration and rotation, but the shown commands only place an expires_at field in JSON and provide a manual cleanup loop; no rotation mechanism is shown.

Skill content
automatic expiration and rotation of shared secrets
Recommendation

Document the actual guarantees clearly, distinguish metadata from enforcement, and add explicit rotation/expiry enforcement if claimed.

#
ASI02: Tool Misuse and Exploitation
Medium
What this means

Secrets could be exposed through local files, logs, shell history, or agent command output if handled carelessly.

Why it was flagged

The Bash workflow writes a raw credential value into a temporary file, including a predictable /tmp path in one command, without showing restrictive permissions or secure handling.

Skill content
cat > /tmp/cred.json <<EOF
{"credential_id":"$(openssl rand -hex 8)","value":"$CRED_VALUE","expires_at":"$EXPIRES_AT"}
EOF
Recommendation

Use restrictive file permissions, randomized private temp files, avoid printing secrets, and prefer a vetted secret-transfer mechanism that does not persist plaintext secrets unnecessarily.

#
ASI06: Memory and Context Poisoning
Medium
What this means

Credential values may remain on disk after use and could be reused, indexed, or exposed later.

Why it was flagged

The workflow persists sent credential files under ~/.pilot/keychain/sent even though the skill says it is not for persistent credential storage.

Skill content
mv /tmp/cred-$cred_id.json ~/.pilot/keychain/sent/
Recommendation

Avoid retaining plaintext sent credentials, enforce automatic deletion for both sent and received files, and document retention behavior clearly.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

The skill may fail or behave differently if required tools are missing or if an unexpected pilotctl binary is on PATH.

Why it was flagged

The skill depends on external tools, and SKILL.md lists jq and openssl in addition to pilotctl, while the registry requirements only declare pilotctl. This is not malicious by itself but users must trust and install the correct tooling.

Skill content
Requires pilot-protocol, pilotctl, jq, and openssl.
Recommendation

Install pilotctl, jq, and openssl from trusted sources and verify the pilotctl binary before using it for credentials.