AgentStack
MCP unreviewed Apache-2.0 Self-run

Offsend

mcp-offsend-offsend · by Offsend

See and fix what AI tools can read. Local-first checks for prompts, file reads, shell, MCP, and agent history — before secrets reach AI context.

No reviews yet
0 installs
2 views
0.0% view→install

Install

$ agentstack add mcp-offsend-offsend

Open-source listing — not yet scanned by AgentStack. Follow the source repository for install instructions.

Security review

⚠ Flagged

1 finding(s); flagged for manual review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures
  • high Pipes remote content directly into a shell (remote code execution).

What it can access

  • Network access Used
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets Used
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Offsend? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

*}• Offsend

See and fix what AI tools can read. One .offsend.yml defines what stays out of AI context — synced to every AI ignore file. Local-first checks for terminals, CI, and macOS — before Claude Code, Codex, Cursor, or Windsurf see your context.

Website · Quick Start · Docs · Check · Extension


.gitignore protects Git. It does not define what AI tools should read.

Offsend is that missing layer: audit what AI tools can see, describe what they must not see in one .offsend.yml, and let offsend sync maintain .cursorignore, .claudeignore, .aiexclude, and the rest for you. Everything runs locally — no cloud account, no upload of file contents for analysis. The CLI is free and open source.

No install yet? Scan a public GitHub repo with Check.

What Offsend does

| Layer | Job | Commands | | --- | --- | --- | | Boundary | Show sensitive paths AI can see; keep them out via .offsend.yml synced to all AI ignore files | show, protect, ignore, sync | | Content | Scan files, staged diffs, or stdin for secrets and custom terms | check | | Runtime | Gate prompts, file reads, shell, MCP calls, and Cursor subagents in the editor; audit local agent history | hook install, history |

Defense-in-depth: ignore files first, then hooks. Hooks do not replace keeping secrets out of the workspace — see [what hooks cover](docs/cli.md#what-hooks-cover).

Quick Start

curl -fsSL https://install.offsend.io/cli | bash
offsend doctor
offsend show

When sensitive paths are exposed:

Scanned: /path/to/project
3 files would be sent to AI tools (2 required, 1 recommended):

✗ Environment files [required]
    Ignore .env and .env.* files.
  - .env

✗ PEM keys [required]
    Ignore PEM key files.
  - server.pem

Recommended onboarding:

offsend init --template node   # .offsend.yml + first sync + baseline check
offsend protect                # promote exposed paths to .offsend.yml, sync AI ignore files
offsend show                   # verify boundary (+ MCP / history hints)
offsend hook install           # git pre-commit + AI editor gates

Rules live in .offsend.yml — commit it and the whole team gets the same boundary. AI ignore files are generated artifacts and stay out of git by default (ignore.commit: false).

Other installs: [CLI docs → Install](docs/cli.md#install) · macOS app: brew install --cask offsend/tap/offsend

Pick your tool

| Tool | Best for | | --- | --- | | [CLI](docs/cli.md) | Repos, git hooks, AI-editor gates, CI (macOS & Linux) | | [macOS app](docs/macos-app.md) | Safe Paste, drag-and-drop prep, watched folders | | Check | One-off scan of a public GitHub repo | | GitHub Action | Same checks on every PR / push | | Extension | Mask secrets in ChatGPT, Claude, Gemini, and similar chats |

CLI essentials

| Command | Purpose | | --- | --- | | offsend show | Sensitive paths visible to AI (+ MCP inventory, agent-history hint) | | offsend protect | Promote exposed paths to .offsend.yml and sync AI ignore files | | offsend sync | Materialize .offsend.yml rules into every AI ignore file | | offsend check | Scan contents (files, --staged, stdin, or editor hook JSON) | | offsend hook install | Git pre-commit + prompt / read / shell / MCP / subagent gates | | offsend history audit | Find secrets already written into local Cursor/Claude transcripts | | offsend doctor | Verify install, hooks, MCP policy, next setup steps |

# CI
- uses: Offsend/ai-hygiene@v1
  with:
    fail-on: block

Full reference: [docs/cli.md](docs/cli.md) · Config: [docs/configuration.md](docs/configuration.md)

Privacy

  • Scanning and audits run on your machine.
  • Offsend does not upload scanned file contents, prompts, clipboard payloads, or findings.
  • Check only analyzes a GitHub repo you choose to scan online.

Details and vulnerability reporting: [docs/faq.md](docs/faq.md) · [SECURITY.md](SECURITY.md).

Docs

| Doc | Description | | --- | --- | | [docs/README.md](docs/README.md) | Documentation index | | [docs/cli.md](docs/cli.md) | Commands, flags, exit codes, AI-editor hooks | | [docs/configuration.md](docs/configuration.md) | .offsend.yml reference (check, ignore, hooks, context) | | [docs/macos-app.md](docs/macos-app.md) | Desktop app, Free vs Pro, App vs CLI | | [docs/faq.md](docs/faq.md) | FAQ | | [.offsend.yml.example](.offsend.yml.example) | Annotated config starter |

Contributing

Bug reports, feature requests, docs improvements, and PRs are welcome.

  • Open an issue
  • Read [SECURITY.md](SECURITY.md) before reporting a vulnerability
  • Keep changes focused and explain the user problem they solve

License

Apache 2.0 — see [LICENSE](LICENSE).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.