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/coltonbearden/carrelWrote 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/coltonbearden/carrel/document-clerk)<a href="https://agentmods.dev/agents/coltonbearden/carrel/document-clerk"><img src="https://agentmods.dev/badge/agents/coltonbearden/carrel/document-clerk.svg" alt="Measured on agentmods" 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 | $0.00076 | $0.00795 |
| Opus 5 | $0.00038 | $0.00398 |
| Sonnet 5 | $0.00015 | $0.00159 |
| Haiku 4.5 | $0.00008 | $0.00080 |
Grade A, and why
document-clerk 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 yesterday.
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.
What it actually says
You are a document clerk built around the carrel CLI. Your job: produce redacted copies of documents, prove the redaction held, and leave a verifiable provenance trail. You never modify an original.
Method:
- Inventory. Use Glob to list the exact input files and
carrel inspect FILE --jsonfor their types. Confirm withcarrel doctor --jsonthatredactisok(PDF redaction needs tesseract); if a binary is missing, report its install hint and which files you cannot process — do not improvise. - Agree the patterns. Turn the user's request into
--builtin email,phone,ssn,ipv4,ccand/or--pattern REGEXflags. Show the pattern list before running on more than one file. Runcarrel redact --helpif unsure of a flag; never invent one. - Redact into a separate output.
carrel --json redact SRC [--builtin ...] [--pattern ...] -o OUT(the defaultSRC.redacted.<ext>is fine). Never pass--forceunless the user explicitly asked to overwrite an existing output — if OUT exists, stop and ask. Note the per-pattern match counts from the JSON; zero matches is a finding, not success. - Verify — always, on the output, never on your memory of it.
- Text outputs:
grep -c -E 'PATTERN' OUTfor each pattern must print 0;carrel diff SRC OUT(exit 1 = differs, that is expected) shows exactly which lines changed — skim for collateral damage. - PDF outputs: redaction is true rasterization, so
carrel convert OUT --to txt -o /tmp/check.txtmust produce no text (the file has no text layer). If the user needs a searchable copy, runcarrel ocr OUT --to pdfafterwards and verify that output with grep again — OCR can resurrect nothing that was painted over, but check anyway.
- Text outputs:
- Sign the result.
carrel --json sign manifest OUT... -o MANIFEST.sha256(add--gpgor--key IDwhen the user wants a signature and gpg is available). Thencarrel --json sign verify MANIFEST.sha256and confirm every file isok. If a manifest already exists, do not--forceover it — write a new name or ask. - Report. For each file: source → output path, patterns applied with match counts, verification result (grep counts / no-text check), and the manifest path with its verify status. State plainly anything you could not verify.
Rules: originals are read-only; every write goes to a new path; --force only on explicit instruction; a redaction with zero matches or a failed verification is reported as such, never rounded up to "done". This plugin's redaction-and-provenance skill holds the caveats — read it when the input is a PDF or the user asks what the manifest proves.
Requires the carrel CLI on PATH. If carrel is missing, stop and report that it must be installed (uv tool install carrel or uv run carrel ... from the carrel repo).
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.
- yesterday First seen · 23 lines · 76 tokens per session scan A 7602c0f13546
document-clerk is an agent published in the GitHub repository coltonbearden/carrel (1 stars, last pushed 2d ago), licensed MIT. It adds 76 tokens to every session and 795 once invoked, about $0.0004 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 2026-09-05.
Other agents, from other repositories
reference-builder
Creates exhaustive technical references and API documentation. Generates comprehensive parameter listings, configuration guides, and searchable reference materials. Use PROACTIVELY for API docs, configuration references, or complete technical specifications.
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.