dchat

PassAudited by ClawScan on May 10, 2026.

Overview

This skill’s network messaging, file transfer, local identity, and message storage are disclosed and aligned with its purpose, but users should treat incoming peer messages and media as untrusted and verify the packaging before use.

This appears aligned with its stated P2P chat purpose. Before installing, verify the package includes the documented wrapper/build files, understand that messages and media are stored under `~/.dchat-clawhub`, and only act on incoming peer messages or send files after confirming the trusted recipient and exact content.

Findings (5)

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.

What this means

If the agent is asked to send the wrong file or to the wrong address, private local content could be shared with a peer.

Why it was flagged

The skill can read a user-selected local file, upload encrypted media through IPFS, and send the decryption information to a peer. This is central to the stated purpose, but it is still a sensitive action.

Skill content
{baseDir}/dchat send-file <address> ./report.pdf  # Send any file (encrypted, via IPFS)
Recommendation

Confirm the recipient address or alias and the exact file path before sending files, images, or audio.

What this means

Anyone who obtains the stored identity material could potentially impersonate the bot on the NKN network.

Why it was flagged

The skill creates and stores a private key that controls the bot’s NKN identity. This is expected for messaging, and the artifacts describe encryption at rest.

Skill content
Seed: 64-char hex Ed25519 private key. Auto-generated at install, saved encrypted to `~/.dchat-clawhub/identity.enc`.
Recommendation

Protect `~/.dchat-clawhub/` and `~/.config/dchat-clawhub/`, and do not share the seed or passkey files.

What this means

The skill may not run as documented, or a user may need to verify/build missing runtime files before trusting the install path.

Why it was flagged

The installer downloads npm production dependencies and expects a prebuilt `dist/cli.js`, while the provided manifest does not show `dist/cli.js` or the `dchat` wrapper referenced in SKILL.md. This looks like a packaging/provenance gap rather than malicious behavior.

Skill content
npm install --omit=dev
...
if [ ! -f dist/cli.js ]; then
  echo "[dchat] ERROR: dist/cli.js not found. The skill package may be corrupt."
...
node dist/cli.js init
Recommendation

Verify the installed package contents, wrapper script, and build artifacts before running the installer or using the skill.

What this means

Old messages, including untrusted text from peers, can persist across sessions and may influence later agent work if read back without context.

Why it was flagged

The skill persists message history locally. This is disclosed and useful for chat history, but stored peer content may later be shown to an agent.

Skill content
History: All messages stored in SQLite at `~/.dchat-clawhub/messages.db`.
Recommendation

Treat message history as untrusted external content and delete or protect the local data directory when appropriate.

What this means

A peer could send misleading instructions, links, or media that appear in the agent’s context if the listener is used.

Why it was flagged

The skill is explicitly designed for bot-to-bot and user-to-bot messaging. The artifacts do not describe an allowlist or trust boundary for incoming peer content, so received messages should be considered untrusted.

Skill content
Send or receive messages to/from another bot or NKN user
Recommendation

Only share the bot address with trusted parties, verify aliases/addresses, and require user approval before taking actions based on incoming messages.