Index of every page on this site: llms.txt
All skills
jeremylongshore avatar

/grammarly-data-safety-guardian

@820dfc8
Run it yourself

Audit and gate a Grammarly document-submission manifest offline before an authorized integration upload. Use when checking supported extensions, file-size and text-length metadata, approval, consent, retention, and approved destination. This skill never accepts or submits document content and is not a replacement for legal, privacy, or security approval. Trigger with "Grammarly submission preflight", "Grammarly document safety", or "Grammarly retention check".

Use this Skill: https://skilld.dev/gh/jeremylongshore/tons-of-skills-marketplace/grammarly-data-safety-guardian

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencessubmission-contract.md

≈842 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Submission manifest contract

The manifest is metadata-only. It never contains document bytes, extracted text, filename, path, upload URL, credential, response body, person identifier, or policy record.

{
  "schema_version": "1",
  "operation": "writing-score",
  "content_sha256": "sha256:0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
  "extension": ".txt",
  "byte_size": 1842,
  "text_character_count": 1260,
  "word_count": 210,
  "classification": "confidential",
  "data_owner_approved": true,
  "consent_confirmed": true,
  "transfer_approved": true,
  "provider_retention_acknowledged": true,
  "api_control_plane_origin": "https://api.grammarly.com",
  "presigned_upload_origin": "https://grammarly-example.s3.amazonaws.com",
  "presigned_upload_origin_approved": true
}

The operation is writing-score, ai-detection, or plagiarism. The extension is .doc, .docx, .odt, .txt, or .rtf; bytes are 1 through 4,194,304. UTF-8 text requires both derived counts, no more than 100,000 characters and at least 30 words. Counts for binary formats are optional because provider extraction cannot be reproduced safely from metadata.

Classification must be public, internal, confidential, or restricted. The guard blocks restricted by default; relabeling is not a bypass. The owning governance process must make any exceptional transfer decision outside this tool. The ownership, consent, transfer, and retention booleans must always be true; the upload-origin approval becomes true only after inspection. api_control_plane_origin must exactly equal the pinned Grammarly API origin. Before inspection, set presigned_upload_origin to JSON null and presigned_upload_origin_approved to false; if every other gate passes, the result is INSPECTION_READY, which authorizes only destination discovery and no document upload. After inspection, presigned_upload_origin must be the exact public HTTPS origin emitted by the evaluator: scheme and host only, never the signed path or query, and the approval flag must be true. The origin must use an S3-shaped hostname under amazonaws.com and the default HTTPS port, a fail-closed local policy pinned from Grammarly's official upload example. The evaluator creates a new job for execution and refuses the upload if its newly issued origin differs. The content digest binds this review to the evaluator dry run but is not an anonymity guarantee.

Origin approval is intentionally an egress-authority decision, not approval of the secret path and query. The authenticated Grammarly control plane issues that short-lived capability; the evaluator keeps it in memory, hashes it for correlation, disables redirects, and sends it only through a TLS connection pinned to an address validated for the approved S3 hostname.

Grammarly documents a 24-hour maximum document-retention boundary and 30-day score availability. provider_retention_acknowledged confirms those provider facts were considered; it does not claim local logs, receipts, backups, or downstream systems use the same retention.

Sources: Writing Score, AI Detection, and Plagiarism Detection. Accessed 2026-09-04.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts2mo3 checks · Risk SAFE
  • Gen Agent Trust Hub2mo

    The skill provides an offline safety auditor for Grammarly document manifests, using a deterministic Python script to enforce size, type, and governance constraints. A low-severity risk of indirect prompt injection exists due to the processing of user-supplied JSON manifests, though the skill implements robust validation and sanitization measures.

  • Socket2mo

    No alerts

  • Snyk2mo

    Risk: LOW · No issues

Signed by skilld at 820dfc8. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 3 months ago
What it can do
Runs commands
version
2.0.0
author
Jeremy Longshore <jeremy@intentsolutions.io>
compatibility
Agent Skills-compatible hosts; requires Python 3.10+
All 1 allowed tools
Bash(python3:*)
Other metadata
tags
[
  "saas",
  "grammarly",
  "data-safety",
  "privacy",
  "governance"
]

README badge

README badge for jeremylongshore/tons-of-skills-marketplace/grammarly-data-safety-guardian