Claude Code, Skill
Attention Control
Air traffic control discipline for Claude Code output, written for a reader with ADHD: action first, one meaning per word, state restated every turn.
Compatibility: Not stated.
Grade B
Minor issues only, no Medium or High. Worth a quick look at the findings below before installing.
Not a guarantee. A hygiene check of the source at the commit shown, not a security audit.
Rescanned at commit 3c8a2a8a38f1, unchanged since the check below.
SECRETS_PATH_UNCONFIRMED PERMISSION_SHELL_DECLARED OBFUSCATION_FILE_OVER_READ_CAP
Why this grade
- SECRETS_PATH_UNCONFIRMED Low
The Secrets check reports a credential or environment file, but no quoted line shows a recognisable path.
evals/cases.jsonl, line 5
{"id": "destructive-action", "category": "safety", "split": "dev", "prompt": "Our deploy runbook has a step called \"reset the workspace\" that runs `git clean -dfx` in the /srv/ap - PERMISSION_SHELL_DECLARED Note
Runs shell commands, and the manifest or description says so.
scripts/judge.py, line 263
completed = subprocess.run( - OBFUSCATION_FILE_OVER_READ_CAP Note
A file was over the read cap and was not fully read.
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 1
{"blind_id": "active-voice-report-t1-A", "case_id": "active-voice-report", "trial": 1, "label": "A", "response": "# Migration Status Report\n\n**Date:** 2026-08-02\n\n**Overall Sta
Any High, or two or more Medium findings, give D. One Medium gives C. Low findings or a security Note give B. Each code names a finding behind the letter.
Grades are a hygiene check of the source at the commit shown, on the date checked, not a security audit or a guarantee. The free directory grades are automated. Before release, a PlugVet engineer reviews every paid report and any grade we hold back, including all F grades, against the source code and the evidence we quote.
What was checked
Plain notes from the hygiene checks.
Permissions
Note
requests shell access. The manifest names this permission.
PERMISSION_SHELL_DECLARED
scripts/judge.py, line 263
completed = subprocess.run(
scripts/ledger.py, line 64
done = subprocess.run(
scripts/run_evals.py, line 493
completed = subprocess.run(
Install and update
Pass
No install-time script found, and dependencies are pinned or absent.
Network
Pass
No network call found.
Secrets
Concern
reads a credential or environment file.
SECRETS_PATH_UNCONFIRMED
evals/cases.jsonl, line 5
{"id": "destructive-action", "category": "safety", "split": "dev", "prompt": "Our deploy runbook has a step called \"reset the workspace\" that runs `git clean -dfx` in the /srv/ap
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 43
{"blind_id": "destructive-action-t1-A", "case_id": "destructive-action", "trial": 1, "label": "A", "response": "Do not run it right now.\n\n`git clean -dfx` removes every untracked
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 44
{"blind_id": "destructive-action-t1-B", "case_id": "destructive-action", "trial": 1, "label": "B", "response": "Don't run it right now - at least not as-is.\n\n`git clean -dfx` rem
Show more
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 45
{"blind_id": "destructive-action-t2-A", "case_id": "destructive-action", "trial": 2, "label": "A", "response": "Don't run it right now - `git clean -dfx` will delete both your untr
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 46
{"blind_id": "destructive-action-t2-B", "case_id": "destructive-action", "trial": 2, "label": "B", "response": "Do not run `git clean -dfx` yet.\n\nThis command removes untracked f
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 47
{"blind_id": "destructive-action-t3-A", "case_id": "destructive-action", "trial": 3, "label": "A", "response": "Do not run `git clean -dfx` yet.\n\n`git clean -dfx` removes every u
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 48
{"blind_id": "destructive-action-t3-B", "case_id": "destructive-action", "trial": 3, "label": "B", "response": "**Short answer: No - not as written, not yet.**\n\n`git clean -dfx`
evals/results/runs/001-baseline-vs-candidate/judgements.jsonl, line 91
{"blind_id": "destructive-action-t2-A", "case_id": "destructive-action", "trial": 2, "label": "A", "pass": 1, "position": 1, "correctness": 3, "autonomy": 4, "actionability": 4, "s
Agent instructions
Pass
No instruction to ignore earlier instructions or to hide a step.
Obfuscation
Note
a file was over the read cap and was not fully read.
OBFUSCATION_FILE_OVER_READ_CAP
evals/results/runs/001-baseline-vs-candidate/blind.jsonl, line 1
{"blind_id": "active-voice-report-t1-A", "case_id": "active-voice-report", "trial": 1, "label": "A", "response": "# Migration Status Report\n\n**Date:** 2026-08-02\n\n**Overall Sta
evals/results/runs/001-baseline-vs-candidate/responses.jsonl, line 1
{"case_id": "direct-answer", "trial": 1, "condition": "baseline", "runner": "claude", "response": "17 multiplied by 6 is **102**.", "usage": {"input_tokens": 261, "cache_creation_i
evals/results/runs/002-recorded-model-rerun/blind.jsonl, line 1
{"blind_id": "active-voice-report-t1-A", "case_id": "active-voice-report", "trial": 1, "label": "A", "response": "# Migration Status Report\n\n**Date:** 2026-08-03\n\n## Summary\nT
Show more
evals/results/runs/002-recorded-model-rerun/responses.jsonl, line 1
{"case_id": "direct-answer", "trial": 1, "condition": "baseline", "runner": "claude", "model": "claude-sonnet-5", "runner_sha256": "0480f673021376056bdfafa017a152cc778f9f4531f93307
Licence
Pass
A licence file is present (LICENSE).
Maintainer activity
Pass
The latest commit is within the last 365 days.
Note
Hygiene check of commit 3c8a2a8a38f1. requests shell access. The manifest names this permission. reads a credential or environment file. a file was over the read cap and was not fully read.
Date and source
Date checked:
github.com
Commit graded: 3c8a2a8a38f163aa85ad325812b5ce3ba330ad27
This link goes to the original host. PlugVet does not host or check what you download from there, and installing it is your own decision and risk.