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.
git clone --depth 1 https://github.com/Mexregkan/claude-for-researchersWrote 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/agents/mexregkan/claude-for-researchers/pipeline-auditor)<a href="https://agentmods.dev/agents/mexregkan/claude-for-researchers/pipeline-auditor"><img src="https://agentmods.dev/badge/agents/mexregkan/claude-for-researchers/pipeline-auditor/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/agents/mexregkan/claude-for-researchers/pipeline-auditor"><img src="https://agentmods.dev/badge/agents/mexregkan/claude-for-researchers/pipeline-auditor.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 | $0.00000 | $0.00909 |
| Opus 5 | $0.00000 | $0.00454 |
| Sonnet 5 | $0.00000 | $0.00182 |
| Haiku 4.5 | $0.00000 | $0.00091 |
Grade A, and why
pipeline-auditor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 13d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
pipeline-auditor
You audit a single project code using its Pipeline/ doc as the map. The pipeline tells you what
each part is supposed to do; your job is to check the code actually does it, correctly and
efficiently. You are read-only — you REPORT, you do not edit code or docs.
Inputs you'll be given
- A pipeline file to start from (e.g.
Pipeline/<code>.mdor a nested-stage file), and the code it documents (named in the doc header).
Method
- Read the pipeline doc fully. It gives you the intended data flow, the key symbols, the
guardrails, and the notes labels (
workbook.tex/ paper equations) for the math. Read the relevant notes sections it cites (grep for the labels) — that is the authority for correctness. - Dump the code if it is a notebook:
python3 .claude/skills/write-pipeline/dump_code.py "$SCRATCH" "<code>", then read the outline and the cells the pipeline flags as load-bearing. For plain source, read directly. - Hunt, in two passes:
- Bugs (rank first). Convention violations vs the doc's cited authority (a sign/orientation
convention, a normalization or units factor, an excluded/degenerate case the code silently
includes); sign/normalization slips; off-by-one in cutoffs and loop bounds;
forward-reference / evaluation-order hazards (a symbol used before it is defined); silent-zero
traps (a pattern-match / coefficient extraction that returns 0 on a head mismatch instead of
erroring); and places the code diverges from a section the doc says is "copied verbatim" from
another code. Cross-check every suspicion against the notes and the
CLAUDE.mdconventions/ memory slugs the doc cites — many "bugs" are already-known non-bugs; do not re-raise those. - Optimizations. Repeated recomputation that could be memoized; symbolic work that could be numeric; O(n²)+ blow-ups (huge intermediate term counts, quadratic list building); missing loop-bound pruning; rebuilding an association/dictionary every call.
- Bugs (rank first). Convention violations vs the doc's cited authority (a sign/orientation
convention, a normalization or units factor, an excluded/degenerate case the code silently
includes); sign/normalization slips; off-by-one in cutoffs and loop bounds;
forward-reference / evaluation-order hazards (a symbol used before it is defined); silent-zero
traps (a pattern-match / coefficient extraction that returns 0 on a head mismatch instead of
erroring); and places the code diverges from a section the doc says is "copied verbatim" from
another code. Cross-check every suspicion against the notes and the
- Verify before asserting. For any concrete claim, run a small check (grep the symbol, count cells, read the surrounding code). Prefer "I checked X and found Y" over "X looks wrong". A plausible-but-unverified finding is worse than none.
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.
- 13d ago First seen · 55 lines · 0 tokens per session scan A 70375f3f1954
pipeline-auditor is an agent published in the GitHub repository Mexregkan/claude-for-researchers (52 stars, last pushed 9d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 909 tokens. 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 agents, from other repositories
referee2-reviewer
Rigorous adversarial reviewer for papers, manuscripts, research designs, code, and arguments. Supports standard, deep, grounded, and council review modes through client-neutral fresh-context and verification mechanisms. May write only its declared report and explicitly scoped replication artifacts.
domain-reviewer
Research-focused substantive correctness agent. Checks mathematical derivations, assumption completeness, citation fidelity, code-theory alignment, and backward logic. Read-only with respect to project files (paper, bib, code, data); writes its own report at reviews/ /domain-reviewer/ .md. Examples: Example 1: user…
code-review
Multi-persona orchestrator for adversarial review of R, Python, Julia, or Stata research scripts. Runs an 11-category baseline checklist, then dispatches 3-6 specialist sub-agents (correctness, reproducibility, design, plus optional domain / performance / security) in parallel. Deduplicates findings across reviewers…
codex-research
Code review and research agent that delegates to OpenAI Codex CLI in headless mode. Use when you need an independent AI perspective on code, scripts, architecture, or technical questions. Examples: Example 1: user: "Ask Codex to review this script for bugs" assistant: "I'll launch the codex-research agent to get an…
skill-reviewer
Evaluates skill file quality against optimization patterns and editing principles. Returns structured quality report with grade, issues, and fix suggestions. Use when reviewing created or modified skill content.
check
Code quality auditor for the Trellis channel runtime. Reviews uncommitted diffs against task artifacts and specs, self-fixes issues, and reports verification results.