tessl-labs/eval-setup
Generate eval scenarios from repo commits, configure multi-agent runs, execute baseline + with-context evals, and compare results — the full setup pipeline before improvement begins
75
94%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Low
Low-risk findings worth noting
Low
Low-risk findings.
1 low severity finding. Worth noting, but not necessarily harmful.
W011: Third-party content exposure detected (indirect prompt injection risk)
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and examines repository commits via `git log --oneline --stat --no-merges -50` or `gh api` (references/phase2-select-commits.md) and then downloads and directly reads user-generated scenario files (`task.md`, `criteria.json`) from `tessl scenario download` (references/phase3-generate-scenarios.md` and phase4), so it ingests untrusted, third-party content that the agent is expected to interpret and that can influence its commit selection, scenario edits, and eval actions.
Where we found it
git repository commits and diffs (user-specified org/repo)
content-type · 3 sites
The plugin instructs the agent to fetch and deeply read commit logs and full diffs from a user-specified repository via git log, git diff, and gh api, then interpret that content to score and select commits — any contributor's commit message or diff content is ingested and influences agent decisions.
references/phase2-select-commits.md
9
Use `git log --oneline --stat --no-merges -50` (or `gh api` if the repo isn't cloned locally) to review recent commits with their diff stats.
37
git diff <hash>~1..<hash>
SKILL.md
58
first scan 50 commits with hard-skip gates
tessl scenario files (task.md, criteria.json)
content-type · 6 sites
The plugin first ingests Tessl-generated scenario content at phase3 (summarizing CLI output), then downloads and reads task.md/criteria.json files at phase4, with the actual read instruction at line 67 — all untrusted generated content influencing agent decisions.
- Audited
- Security analysis