AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
MCP verified Apache-2.0 Self-run

Writ

mcp-sirrobot01-writ · by sirrobot01

Agent orchestration as contract enforcement. Author a writ; mandate, grants, bounds, proof of done and a runtime governs every action

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

Install

$ agentstack add mcp-sirrobot01-writ

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • 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.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/mcp-sirrobot01-writ)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude DesktopCursorWindsurf

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Writ? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Writ

Writ is a language and runtime for bounded agent authority.

A writ file states what an agent can do and how much it can spend. It also states what the agent must not change. The registrar checks each action against these terms. It writes each action, check, and result to a record.

Writ file

A writ can contain these terms:

| Term | Purpose | | --- | --- | | mandate | States the goal. | | grants | Give permission for actions and resources. | | bounds | Limit iterations, cost, time, and changes. | | invariants | State conditions that must remain true. | | obligations | State checks that must pass before satisfaction. | | satisfaction | States the proof that the work is complete. | | remedies | State the response to a breach. |

Example:

writ: bug-fixer
mandate: "The failing tests pass without changes to the tests."

grants:
  - read:  { paths: ["**/*"] }
  - shell: { commands: ["pytest *"] }
  - edit:  { paths: ["src/**"] }

bounds:
  iterations: 40
  cost: $2.50
  wall_clock: 20m
  edits:
    max_changed_lines: 200

invariants:
  - layer: letter
    untouched: "tests/**"

satisfaction:
  all:
    - layer: letter
      run: "pytest -x -q"
      exit: 0

remedies:
  on_bound_breach: { do: halt, inform: issuer }

Check layers

Each clause names the party that makes its check.

| Layer | Check | | --- | --- | | letter | The registrar makes a deterministic check. | | judgment | A separate judge model evaluates the clause. | | assent | A person makes the decision. |

The record keeps the layer with each check. A judgment is not shown as a deterministic result.

Run modes

Writ supports three run modes.

| Mode | Command | Coverage | | --- | --- | --- | | Embedded | writ issue | Model turns, tool actions, effects, bounds, invariants, satisfaction, and remedies. | | Model gateway | writ gate | Model iterations, model cost, and tool names that cross the model API. | | MCP sidecar | writ proxy -- | Tool names and arguments that cross the MCP boundary. |

The model gateway does not run tools. It cannot verify tool effects.

The MCP sidecar cannot see model iterations or model cost. It cannot verify effects outside the MCP calls.

Install

Install Writ with Homebrew:

brew install sirrobot01/tap/writ

You can also download a binary from the releases page.

Check the installation:

writ --version

Start a run

  1. Create the engagement configuration.

``sh writ configure ``

  1. Create a writ file.

``sh writ init bug-fixer ``

  1. Add the required terms to bug-fixer.yaml.
  1. Check the file.

``sh writ check bug-fixer.yaml --explain ``

  1. Issue the writ.

``sh writ issue bug-fixer.yaml ``

  1. Read the record.

``sh writ record bug-fixer ``

Commands

| Command | Purpose | | --- | --- | | writ init | Create a writ file. | | writ check | Validate and explain a writ file. | | writ configure | Configure holders, judges, the sandbox, and assent. | | writ issue | Run an agent under a writ. | | writ gate | Govern an external agent at the model API boundary. | | writ proxy | Govern an MCP server at the tool boundary. | | writ record | Show the record of a run. | | writ amend | Change the terms of a writ. |

Development

Run all Go checks:

task check

Build the command:

task build

Read [CONTRIBUTING.md](./CONTRIBUTING.md) before you send a change.

License

Writ uses the [Apache License 2.0](./LICENSE).

Brand files are in [blog/brand/](./blog/brand).

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.