Skip to main content
Glama

Aegis

Fail-closed control layer for coding agents. MCP tool, not a chat app.

Hooks inject. Tools execute. File-exists is not a receipt. You do not certify.

Tools

Tool

Job

land_inject

Force LAND context this turn (truth + act)

no_fake_done

BLOCK a done-word with no live proof

trust_freeze

Freeze the seat after BLOCK

decide

Run the scar engine on a prompt

Related MCP server: Sensory-Grounding MCP

Cursor

.cursor/mcp.json:

{
  "mcpServers": {
    "aegis": {
      "command": "npx",
      "args": ["-y", "github:Viss007/aegis"]
    }
  }
}

Local clone:

{
  "mcpServers": {
    "aegis": {
      "command": "node",
      "args": ["bin/aegis-mcp.mjs"]
    }
  }
}

Hook (inject, not a tool):

{
  "version": 1,
  "hooks": {
    "beforeSubmitPrompt": [{ "command": "node hooks/land.cjs" }]
  }
}

Scar

Same file on disk. No live proof.

  • inject on → refuse. file-exists side door blocked.

  • inject off → liar says DONE.

HTTP

POST /mcp JSON-RPC. Same methods as stdio (initialize, tools/list, tools/call).

Available Tools

4 tools
decideC

Run the LAND scar engine on a user prompt. Inject on = file-exists side door blocked.

Inputs: prompt (required), claim, inject, file_exists, +1 more.

land_injectC

Force LAND context this turn. Returns additional_context the model must see before the prompt. Hooks inject. Tools do not.

Inputs: no named parameters.

no_fake_doneA

Fail-closed done gate. BLOCK if a done-word is claimed without live proof. file-exists is not a receipt. You do not certify.

Inputs: claim (required), file_exists, live_verified.

trust_freezeC

Freeze the seat after a no_fake_done BLOCK. Second catch. Fail-closed.

Inputs: reason, source.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.7/5.0

    Scored across 4 tools

    Disambiguation2/5

    land_inject and decide both concern LAND context injection, making their boundary unclear. no_fake_done and trust_freeze are sequentially coupled but their distinct roles are not obvious from names alone, so an agent could easily misselect.

    Naming Consistency3/5

    All names use snake_case, but the patterns are mixed: land_inject is verb_noun, no_fake_done is a negation phrase, trust_freeze is noun_verb, and decide is a single verb. It is readable but not predictable.

    Tool Count4/5

    Four tools is a reasonable size for a specialized guardrail server. However, the overlap between land_inject and decide suggests some consolidation might be warranted.

    Completeness2/5

    The inferred domain is agent safety and enforcement, but there is no tool to unfreeze trust, check current state, or submit proof/receipts. These gaps could lead to deadlocks or agent failures.

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      F
      maintenance
      Structured workspace runtime for long-running coding agents, providing controlled workspace capabilities with task state, snapshots, checkpoints, drift detection, verification evidence, audit logs, and structured handoff.
      20
      MIT
    • A
      license
      C
      quality
      C
      maintenance
      A fail-closed preflight, approval, evidence, and verification runtime for agents, preventing unsupported output from being treated as verified completion.
      3
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      A personal agent control harness that governs AI agents by checking every consequential action against user-defined rules, logging all actions to a tamper-evident audit chain, and providing a global kill switch. It runs as an MCP server sidecar to coding agents or standalone, keeping users in charge of what their agents can do.
      8
      AGPL 3.0