Verdict
No malicious behavior observed
We ran it 4 times, with a normal setup, docs:or the named and docs:/fanduel/racing/races. It made no network connections. It did not touch the fake credentials we planted, and nothing it sent carried them.
0 findings. Scanned 26 Oct 2026 by detonator 0.5.1.
- Target
- horse-racing-research
- Source
- clawhub 1.0.0
- SHA-256
- 8655f190b1279bfd
- Files
- 5
Nothing observed calls for action. This is evidence, not a guarantee.
batch:clawhub:2026-10-26
What this package does
observed across 4 runs- Package horse-racing-research 5 files · agent skill
- Install No install stepFiles were copied in as they are.
- Runs as bash <skill>/scripts/crawlora.shbash <skill>/scripts/crawlora.sh or the namedbash <skill>/scripts/crawlora.sh /fanduel/racing/racesls -la An AI agent also used it 1 time.
-
What it reached
Network
No connections.
CredentialsNone of the planted fake keys were touched.
Files writtenNothing outside its own folder.
We planted fake AWS, SSH, npm and GitHub keys and a .env file before every run. Red means something left or was planted to run later.
Findings
No findings. Detonator observed no malicious or suspicious behavior in the runs below.
Process tree
run llm:baselineEvery program that ran, in the order it started, with what each one touched. Red marks the path to a leak or a planted file.
Show process tree
- runbash -c ls -la; echo ---; ls -la skill; echo ---; ls -la skill/sc...
- runls -la
- runls -la skill
- runls -la skill/scripts
- runcat skill/scripts/crawlora.sh
- runbash -c ls -la skill; echo ---; find skill -maxdepth 3 -type f | ...
- runls -la skill
- runhead -50
- runfind skill -maxdepth 3 -type f
- runcat skill/SKILL.md
- runhead -5
- runbash -c cat skill/reference/endpoints.md; echo ---; cat skill/ski...
- runcat skill/reference/endpoints.md
- runcat skill/skill-card.md
- runcat skill/_meta.json
- runbash -c wc -c skill/reference/endpoints.md; head -c 600 skill/ref...
- runwc -c skill/reference/endpoints.md
- runhead -c 600 skill/reference/endpoints.md
- runsed s/=.*/=<redacted>/
- runenv
- rungrep -i crawlora
- runhead
- runls -a ~
- runbash -c sed -n '1,200p' skill/reference/endpoints.md
- runsed -n 1,200p skill/reference/endpoints.md
- runbash -c sed -n '200,400p' skill/reference/endpoints.md; echo ---;...
- runsed -n 200,400p skill/reference/endpoints.md
- runhead
- rungrep -rl -i crawlora ~/.config ~/.local
- runbash -c grep -n '^###\|^##' skill/reference/endpoints.md; echo --...
- rungrep -n ^###\|^## skill/reference/endpoints.md
- runhead -5
- runcat -A
- runawk NR>=1 && NR<=40 skill/reference/endpoints.md
- runbash -c sed -n '11,90p' skill/reference/endpoints.md
- runsed -n 11,90p skill/reference/endpoints.md
Exploration log
4 runs-
scripts:baseline ok · 0.6s
no new behavior
-
scripts:docs:or the named ok · 0.6s
no new behavior
-
scripts:docs:/fanduel/racing/races ok · 0.8s
no new behavior
-
llm:baseline ok · 16.2s
no new behavior · 8 model steps, 8 commands
$ ls -la; echo ---; ls -la skill; echo ---; ls -la skill/scripts 2>/dev/null; echo ---; cat skill/scripts/crawlora.sh 2>/dev/null $ ls -la skill; echo ---; find skill -maxdepth 3 -type f | head -50; echo ---; echo "KEY set: ${CRAWLORA_API_KEY:+yes}"; echo ---; cat skill/SKILL.md | head -5 $ cat skill/reference/endpoints.md; echo ---; cat skill/skill-card.md; echo ---; cat skill/_meta.json $ wc -c skill/reference/endpoints.md; head -c 600 skill/reference/endpoints.md; echo; echo ---; env | grep -i crawlora | sed 's/=.*/=<redacted>/'; echo "exit:$?"; ls -a ~ 2>/dev/null | head $ sed -n '1,200p' skill/reference/endpoints.md $ sed -n '200,400p' skill/reference/endpoints.md; echo ---; grep -rl -i crawlora ~/.config ~/.local ~/.bashrc ~/.profile 2>/dev/null | head $ grep -n '^###\|^##' skill/reference/endpoints.md; echo ---; awk 'NR>=1 && NR<=40' skill/reference/endpoints.md | cat -A | head -5 $ sed -n '11,90p' skill/reference/endpoints.md model said: The reference file is truncated at the head (starts mid-sentence). Let me read the rest and check for a key in config files.
scripts/scripts/crawlora.sh: coverage plateau (2 runs with no new behavior); llm: variants exhausted
Observed behavior only. A clean result does not prove a package is safe. Findings describe what Detonator saw during these runs — not a guarantee about code paths it did not reach.