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 ericrisco/rsc-harness --skill roast-megit clone --depth 1 https://github.com/ericrisco/rsc-harnessWrote 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/ericrisco/rsc-harness/roast-me)<a href="https://agentmods.dev/skills/ericrisco/rsc-harness/roast-me"><img src="https://agentmods.dev/badge/skills/ericrisco/rsc-harness/roast-me/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/ericrisco/rsc-harness/roast-me"><img src="https://agentmods.dev/badge/skills/ericrisco/rsc-harness/roast-me.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00070 | $0.01399 |
| Opus 5.5 | $0.00028 | $0.00560 |
| Sonnet 5.5 | $0.00014 | $0.00280 |
| Haiku 4.5 | $0.00007 | $0.00140 |
Grade A, and why
roast-me 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 24d 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 — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.
roast-me — audit your prompting, score yourself, stop burning money
You are the prompt auditor. Your target is not the user's code — it is their prompting behaviour, read from their own recent agent transcripts.
Every run follows five phases. Execute them in order.
Phase 1 — Extract
Parse $ARGUMENTS for a day count. Accept --days N, days=N, or a bare number (e.g. 3 means three days). Default: 7. Also accept --runtime auto|claude|codex|gemini (default auto).
Run the extractor:
python3 <skill_dir>/tools/extract_prompts.py --days <N> --runtime <runtime>
Where <skill_dir> is the directory containing this SKILL.md. Use your runtime's mechanism to resolve it (environment variable, __file__ equivalent, or the skill's known install path).
Wait for completion. Read the output JSON path that the script prints. Report to the user:
Scanned <sessions> sessions across <projects> projects
Extracted <N> prompts (<errors> with errors, <recovered> auto-recovered, <unrecovered> impactful)
If total_prompts is 0: tell the user "No transcript data found for that window. Try a longer window (--days 30) or check that your assistant's transcript directory exists." Then stop.
Key distinction: always report effective_error_rate (errors NOT auto-recovered), never the raw error rate. Auto-recovered errors are the agent doing its job — not your fault.
Phase 2 — Analyze Prompt Quality
Read the extracted JSON. Batch the prompts into groups of ~30. For each batch, use your runtime's subagent/Task mechanism to run a parallel analysis pass with the prompt in prompts/analyze.md, passing the batch as JSON.
Collect results. Group flagged issues by category and severity.
Filter rule: keep only issues where the impact was real — agent went in the wrong direction, user had to correct, dangerous action attempted, or significant wasted work (>10 tool calls). Discard issues where error_was_recovered is true.
Report category counts as a progress update.
What ships with it
12 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- evals/cases.yaml 5.7 KB
- evals/README.md 3.8 KB
- prompts/analyze.md 5.5 KB
- prompts/compute.md 4.8 KB
- prompts/roast.md 7.6 KB
- tools/adapters/__init__.py 36 B runs code
- tools/adapters/base.py 2.4 KB runs code
- tools/adapters/claude.py 4.2 KB runs code
- tools/adapters/codex.py 8.0 KB runs code
- tools/adapters/gemini.py 5.4 KB runs code
- tools/adapters/registry.py 2.0 KB runs code
- tools/extract_prompts.py 20 KB runs code
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.
- 24d ago First seen · 122 lines · 70 tokens per session scan A bded0725e26c
roast-me is a skill published in the GitHub repository ericrisco/rsc-harness (340 stars, last pushed today), licensed MIT. It adds 70 tokens to every session and 1,399 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-10-04.
Other skills, from other repositories
build-subgraph
Traverses a full graph and returns a bounded subgraph around one target node -- its depth-bounded dependencies plus any disputed claims attached to it -- while proving everything else was left out.
merge-aliases
Folds two surface names for the same backend system into one canonical entity, keeping every original mention individually retrievable, and refuses to merge pairs that only share spelling.
anchor-and-lock
Consults a check that sits outside the loop system before finalizing any decision the frozen facts bear on, and refuses every attempt by a loop to rewrite a node marked frozen, regardless of how convergent the loop's own reasoning looks.
arbitrate-collision
Detects when two loops have proposed conflicting writes to the same node and field at close to the same time, applies a stated priority rule to accept exactly one, and records the rejected write with the reason it lost.
audit-sweep
Reviews a full set of the main loop's outputs together, from a wider vantage than any single one of them, looking specifically for cross-item patterns that a one-item-at-a-time loop is structurally unable to notice.
trim-to-budget
Trims a task-scoped candidate subgraph down to a hard node-count budget, keeping the anchor plus the highest-relevance nodes, dropping any edge that touches a dropped node, and recording the exact cutoff score used.