CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

30.8 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 1 flagged

Securityscore 88 · 1 finding

MEDIUM"Never tell the user" non-disclosure imperative in the skillSS-SKILL-INJECT-IMPERATIVE-01 · Prompt injection · README.md

MEDIUM — it fires on intent; whether the agent honors the non-disclosure imperative depends on the host model.

Why it matters

A bulleted imperative like {match} tells the agent to never reveal, disclose, or mention something to the user. Used adversarially it can instruct the agent to hide its tool calls or lie about what it did — stripping the transparency a user relies on to trust the agent.

The exact value spotted

excerptREADME.md· markdown

840- Never commit `update_hr_contacts.py` to Git (contains real recruiter names/LinkedIn URLs)

841- Keep your Gmail App Password secure

842- Don't share your Excel tracker if it contains private data

843

844## Acknowledgements

Occurrences

1 occurrence · at L842

How to fix

Remove the non-disclosure imperative, or rescope it so it limits output format, not honesty to the user.

Delete any "never reveal / never tell the user" line aimed at the model's own behavior.
Keep legitimate confidentiality rules about external data (e.g. "do not echo API keys"), which are user-protective, not user-deceiving.

Framework references

OWASPLLM01 ATLASAML.T0051

Trace & refs

ruleSS-SKILL-INJECT-IMPERATIVE-01sha256d9839cdc44f5a183…rubric 365aacaView on GitHub

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

.claude/skills/new-job/SKILL.md · 1 file

.claude/skills/new-job/SKILL.md1.8 KB · Markdown

name: new-job

description: This skill should be used when the user wants to add a new company or job to their Excel tracker (List.xlsx). Triggers on phrases like "add [company] to tracker", "I applied to [company]", "track [company]", "save [company] for later", "add [company] [role]", "new application at [company]".

Add New Job to Tracker

Add a new row to List.xlsx (Sheet1) with the bundled script. Do not write ad hoc openpyxl code.

Column layout

A: Company name
B: Role title
C: Role link (job posting URL, without tracking redirects; /thanks pages are fine)
D: Status
E: HR contact (leave empty; use the update-hr skill)
F: Comments (optional)

Status values

done = applied (default)
In progress = ONLY when a callback/response was received
Rejected = only when logging a rejection for a job that was never tracked

Steps

Get the company name (required), role title and job URL. If the user only pasted a link, fetch the page (or the ATS public API: Ashby/Lever/Greenhouse) to get company and role.
Run from the project root (claude-job-agent/):

   python .claude/skills/new-job/scripts/add_job.py --company "Company" --role "Role" --url "https://..." [--status done] [--comment "..."]

Handle the result:

ADDED: row N (exit 0) → confirm company, role, status and row to the user.
DUPLICATE (exit 2) → the same job URL is already tracked; tell the user the existing row, do not add.
COMPANY EXISTS (exit 3) → show the existing rows. If it's a different role, re-run with --force.
ERROR (exit 1) → e.g. Excel has the file open; ask the user to close it and retry.

The script takes the shared tracker lock, backs up List.xlsx to backups/ before writing, and matches URLs ignoring www, query strings, trailing slashes and /thanks or /application suffixes.

new-job— agent skill

Is new-job safe to install?

Add New Job to Tracker

Column layout

Status values

Steps

Audit the pieces. Scan the whole. Decide.