Agent Economy ReportAgent Economy ReportCAUTION version 1.2.2 · audited 2026-09-17
This package presents itself as a Clawdbot documentation helper. It mostly contains markdown guidance and simple shell scripts that print messages rather than actually downloading, searching, or tracking documentation. No code was found that reads or transmits credentials, installs software, persists on the machine, or contacts hidden destinations. The main concerns are that the advertised search, fetch, index, and tracking functions are only stubs, and one configuration example shows a gateway binding to all network interfaces. No malicious behavior was found, but users should review the scripts and generated configuration before relying on them.
| Declared purpose | Clawdbot documentation expert with decision tree navigation, search scripts, doc fetching, version tracking, and config snippets for all Clawdbot features |
| Observed behavior | The skill instructs the agent to choose documentation paths, run helper shell scripts, and use browser or configuration snippets. The shell scripts use bash but only echo status messages or user-supplied values; they do not actually fetch, search, index, cache, or snapshot anything. External references are docs.clawd.bot and a clawhub.ai profile link. Snippets include Discord and Telegram token environment-variable placeholders and a gateway host 0.0.0.0 example. |
| Verdict, rules only | CAUTION · reads credential-like environment variables: DISCORD_TOKEN, TELEGRAM_TOKEN |
| Verdict, AI | CAUTION |
| Final verdict | CAUTION (never better than either pass) |
| capability | vs purpose | detail |
|---|---|---|
| shell | needed | SKILL.md directs the agent to run scripts such as ./scripts/sitemap.sh and ./scripts/search.sh; each script has a bash shebang and uses echo/case statements. No destructive shell command was found. |
| network | needed | The declared purpose is fetching and searching documentation from docs.clawd.bot, but the included scripts only echo URLs and do not make network requests. |
| browser | needed | SKILL.md line 128 says to fetch documentation with the script or use a browser. |
| credentials | exceeds purpose | snippets/common-configs.md lines 9 and 23 contain ${DISCORD_TOKEN} and ${TELEGRAM_TOKEN} placeholders. No script reads or transmits these variables. |
| severity | finding |
|---|---|
| info | Declared purpose This is the stated purpose of the package. The included scripts do not fully implement the search, fetching, indexing, and tracking features described here. SKILL.md:3 description: Clawdbot documentation expert with decision tree navigation, search scripts, doc fetching, version tracking, and config snippets for all Clawdbot f |
| info | Shell helper script usage The skill instructs the agent to run local shell scripts. This requires shell capabilities in the host agent. SKILL.md:50 ./scripts/sitemap.sh # Show all docs by category |
| medium | Doc fetch script is a stub Despite the command name and advertised doc-fetching, this script only prints the URL and does not download or fetch the document. scripts/fetch-doc.sh:7 echo "Fetching: https://docs.clawd.bot/$1" |
| medium | Search script is a stub The advertised keyword search does not search an index or documentation; it only echoes the provided keyword. scripts/search.sh:7 echo "Searching docs for: $1" |
| medium | Full-text index download is a stub SKILL.md says this requires qmd and can download all docs, but the script only prints a message and never invokes qmd or downloads anything. scripts/build-index.sh:5 echo "Downloading all docs..." |
| low | Version tracking is a stub The snapshot, list, and since commands only print messages; they do not save or compare any state. scripts/track-changes.sh:5 echo "Saving current state..." |
| info | Sitemap script is a local category lister The script prints hard-coded category names and does not fetch or parse a sitemap from docs.clawd.bot. scripts/sitemap.sh:22 echo "📁 /$cat/" |
| medium | Example binds gateway to all interfaces If copied into a real configuration, this would expose the gateway on all network interfaces. skill-card.md lines 25-27 acknowledge this risk and recommend using 127.0.0.1 or adding authentication/firewall controls. snippets/common-configs.md:41 "host": "0.0.0.0", |
| info | Discord token placeholder in config snippet This uses an environment variable placeholder for a sensitive credential. No script reads or transmits the variable; it is an example configuration. snippets/common-configs.md:9 "token": "${DISCORD_TOKEN}", |
| info | Publisher acknowledges gateway exposure risk The package itself documents the risk associated with the 0.0.0.0 example, so the risky config is not hidden. skill-card.md:25 Risk: Copying the gateway example with host 0.0.0.0 can expose a local service if network controls are not in place. |
0 AI finding(s) were dropped because their file, line or quote did not match the package.
| domain | service rating | where |
|---|---|---|
| clawhub.ai | not in the service index | skill-card.md:9 |
| docs.clawd.bot | not in the service index | SKILL.md:138 |
| Binaries invoked | bash |
| Environment variables read | CATEGORIES, DAYS, DISCORD_TOKEN, TELEGRAM_TOKEN |
| Hard-coded wallets | none |
| Pipes a download to a shell | 0 |
| eval / exec / subprocess | 0 |
| base64 blobs | 0 |
| File writes | 2 |
| Persistence | 0 |
| file | lines | sha256 |
|---|---|---|
SKILL.md | 166 | 114fb5408ffe9f0b… |
_meta.json | 6 | 89313a792b1f69fa… |
package.json | 9 | 4a3e6c6090ac9832… |
scripts/build-index.sh | 17 | 0d35e8760b37dd2c… |
scripts/cache.sh | 13 | a86f318b3a3f1116… |
scripts/fetch-doc.sh | 7 | fe70080e81b93565… |
scripts/recent.sh | 5 | a53e22fe1818710c… |
scripts/search.sh | 8 | 520eb36668567368… |
scripts/sitemap.sh | 23 | 92fe84e5597820b5… |
scripts/track-changes.sh | 16 | d80f6bd0df132d58… |
skill-card.md | 51 | 3524160718349bd5… |
snippets/common-configs.md | 69 | 4140e9c3622efa67… |
For agents
JSON: https://agenteconomy.report/k/clawddocs.audit.json · badge: https://agenteconomy.report/k/clawddocs.audited.svg ·
skill rating: /k/clawddocs · commission an audit of any skill: US$ 29 per version.
The complete published package of this exact version was downloaded from the registry and read statically; nothing was executed. A deterministic pass extracts network destinations, binaries, environment variables, writes, install commands, obfuscation markers, subprocesses and wallets, each with file and line. An AI then reads the whole package with those facts and writes the summary, the capabilities and the findings under a strict schema; every finding it produces must cite a file, a line and the exact text, or it is dropped. The final verdict is the worst of the two passes. The audit does not change the skill's trust tier (policy); the author may respond through the dispute channel and the response is published here. Commissioned by: the Agent Economy Report (free program: the 150 most downloaded skills, September 2026).