Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add VonTerraProject501c3/slushpile --skill adversarial-reviewgit clone --depth 1 https://github.com/VonTerraProject501c3/slushpileWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/vonterraproject501c3/slushpile/adversarial-review)<a href="https://agentmods.dev/skills/vonterraproject501c3/slushpile/adversarial-review"><img src="https://agentmods.dev/badge/skills/vonterraproject501c3/slushpile/adversarial-review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/vonterraproject501c3/slushpile/adversarial-review"><img src="https://agentmods.dev/badge/skills/vonterraproject501c3/slushpile/adversarial-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00056 | $0.03597 |
| Opus 5 | $0.00028 | $0.01799 |
| Sonnet 5 | $0.00011 | $0.00719 |
| Haiku 4.5 | $0.00006 | $0.00360 |
Grade A, and why
adversarial-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 257 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adversarial Review
Seven agents try to reject the application before a recruiter gets the chance.
The output is deliberately not a single yes or no. The same materials convert at wildly different rates through a cold portal submission than through a referral, and a review that collapses those into one verdict is telling the user something false in a format that sounds authoritative.
Announce at start: "Running adversarial review for $ROLE. Extracting materials, launching five specialists in parallel."
Arguments:
$1— path to a role folder containing at minimum a resume andjob_description.md
Example:
/slushpile:adversarial-review applications/Acme/Engineering/Staff-SRE
The Pipeline
| # | Agent | Model | When |
|---|---|---|---|
| 1 | slushpile-triage-screener |
sonnet | parallel |
| 2 | slushpile-requirements-analyst |
sonnet | parallel |
| 3 | slushpile-ats-simulator |
sonnet | parallel |
| 4 | slushpile-fatigued-reader |
sonnet | parallel |
| 5 | slushpile-pool-analyst |
opus | parallel |
| 6 | slushpile-hiring-manager |
opus | after 1-5 |
| 7 | slushpile-contrarian |
opus | after 6 |
Stages 1 through 5 run concurrently. Dispatch them in a single message with five tool calls.
On a harness without subagent dispatch: run the seven personas sequentially in one context, reading each agent definition from the plugin's agents/ directory and adopting it in turn. Write each report out before starting the next, and do not let a later persona see an earlier one's conclusion except where the pipeline says it should — the hiring manager gets all five specialist reports, the contrarian gets everything, and the five specialists get nothing from each other. Contamination between the parallel stages is the main thing that degrades in a sequential run, and it degrades quietly.
Why It Is Shaped This Way
Four failure modes drove the design. Each maps to a stage.
Single-perspective sycophancy. Every reviewer in a naive pipeline works for the candidate. None of them models the queue. The pool analyst exists to force comparative reasoning: not "are these materials good" but "are they better than the other seventy applications this week."
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 257 lines · 56 tokens per session scan A ff5b5ab587b1
adversarial-review is a skill published in the GitHub repository VonTerraProject501c3/slushpile (15 stars, last pushed 25d ago), licensed MIT. It adds 56 tokens to every session and 3,597 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
create-diagram
Create architecture diagrams from a local codebase or system description. Use when the user asks to diagram a repo, map system architecture, show component relationships, create drill-down views, or explain a system visually with tld.
resume
Generate a tailored resume AND cover letter for a job description via the native four-role Resume Team, score both against ATS and HR rubrics, create DOCX files, and update the tracker. Use when the user pastes a job description and wants a complete application package (resume plus cover letter) with dual scoring and…
writing-coach
Human-voice writing coach that rewrites resumes and cover letters for brevity, burstiness, plain language, and authentic impact, and blocks AI-sounding prose. Use when the user wants to improve writing quality, fix robotic or generic AI-sounding text, cut fluff, strengthen bullets and summaries, or pass the…
cover-letter
Create a compelling one-page, human-voice cover letter for a job description and generate the final DOCX. Use when the user wants a cover letter only (no resume), pastes a JD and asks for a letter, or needs a letter to accompany an already-tailored resume. Runs the mandatory humanvoiceaudit before producing the DOCX.
find-jobs
Search live job boards for roles that match the master resume, then score and rank them by ATS and HR fit. Use when the user wants to discover or find jobs, search openings by title or location, wants remote roles, or asks which live listings best match their background. Uses the discoverjobs MCP tool and never…
job-fit
Run the deterministic, digest-bound candidate-fit gate that scores the configured master resume against an exact job description before any resume tailoring. Use when the user wants a GO/NO-GO fit check on a JD, asks "should I apply", or before starting resume work, to confirm score >= 70 with zero hard knockouts.…