MCP Tools Reference

The Appgrammar MCP server exposes 31 tools via the Model Context Protocol. These tools allow AI coding agents to create, execute, transform, compose, and analyze appgrammars programmatically.

Connection

Add the following to your MCP client configuration (e.g. claude_desktop_config.json):

{
  "mcpServers": {
    "appgrammar": {
      "url": "https://api.appgrammar.com/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_MCP_TOKEN>"
      }
    }
  }
}

Generate your MCP token from the MCP Setup page.

Tools by Axiom

Axiom I — Compression

"Knowledge compresses. Execution expands."

Tool	Description	Multi-turn
appgrammar_create	Create a new appgrammar blueprint from a natural-language description.	Yes
appgrammar_create_submit	Submit LLM output for the current step in the appgrammar creation pipeline.	Yes
appgrammar_decompile	Reverse-engineer an existing codebase into an executable appgrammar.	Yes
appgrammar_decompile_submit	Submit LLM output for the current step in the decompilation pipeline.	Yes
appgrammar_holographic_doc	Generate a holographic documentation artifact from an appgrammar.	Yes

Axiom II — Context

"Agent quality degrades monotonically with context size."

Tool	Description	Multi-turn
appgrammar_start	Start executing an appgrammar blueprint and receive the first step.	No
appgrammar_next	Get the next step in an appgrammar blueprint execution.	No
appgrammar_status	Get the status of an appgrammar execution including steps completed and total steps.	No
appgrammar_swarm_partition	Partition an appgrammar into parallel execution swarms.	Yes
appgrammar_semantic_rollback	Create a semantic rollback point for an appgrammar.	Yes

Axiom III — Governance

"Ungoverned agents are liabilities."

Tool	Description	Multi-turn
appgrammar_inoculation	Inoculate an appgrammar against common failure patterns.	Yes
appgrammar_phenotyping	Analyze the phenotype of an appgrammar — its observable characteristics, behaviors, and output patterns.	Yes

Axiom IV — Architecture

"Application structure is composable, forkable, and heritable."

Tool	Description	Multi-turn
appgrammar_breed	Breed two appgrammars by analyzing module graphs and merging selected modules from each parent.	Yes
appgrammar_fork	Fork a public appgrammar to create your own independent copy.	No
appgrammar_splice	Splice two appgrammars together at compatible module boundaries.	Yes
appgrammar_shatter	Shatter an appgrammar into its atomic module components.	Yes
appgrammar_weave	Weave multiple appgrammar modules into a unified blueprint.	Yes

Axiom V — Ecosystem

"When intent is legible, the ecosystem becomes navigable."

Tool	Description	Multi-turn
appgrammar_search	Search the appgrammar registry for public blueprints.	No
appgrammar_broadcast	Send a broadcast notification to all subscribers, forkers, and team members of an appgrammar.	No
appgrammar_topological_search	Search the appgrammar registry using topological similarity.	Yes
appgrammar_upstream_shield	Apply an upstream shield to an appgrammar to protect against breaking dependency changes.	Yes

Axiom VI — Transformation

"Intent is invariant under transformation."

Tool	Description	Multi-turn
appgrammar_transmute	Transmute an appgrammar from one technology stack to another while preserving architectural intent.	Yes
appgrammar_distill	Distill an appgrammar to its essential components.	Yes
appgrammar_amplify	Amplify an appgrammar by expanding its modules with additional detail, tests, and documentation steps.	Yes
appgrammar_style_transfer	Transfer the coding style and conventions from one appgrammar onto another while preserving functionality.	Yes

Axiom VII — Algebra

"Structure admits a closed algebra."

Tool	Description	Multi-turn
appgrammar_subtract	Subtract one appgrammar from another, removing modules and steps present in the second from the first.	Yes
appgrammar_intersect	Compute the intersection of multiple appgrammars, returning only shared modules and steps.	Yes
appgrammar_difference	Compute the symmetric difference between two appgrammars, highlighting unique modules in each.	Yes
appgrammar_fuzz	Fuzz an appgrammar by introducing controlled variations to test robustness and discover edge cases.	Yes
appgrammar_update	Update an existing appgrammar to reflect codebase changes.	Yes

Shared

Infrastructure shared across multi-turn tool operations.

Tool	Description	Multi-turn
appgrammar_tool_submit	Submit data for a multi-turn tool operation.	Yes

Multi-turn Protocol

Many tools use a multi-turn protocol where the initial call returns a prompt and token rather than a final result. The workflow is:

1. Call the tool — receive a prompt, expected output schema, and a single-use token
2. Run the prompt on your LLM with the provided schema constraints
3. Submit the result via appgrammar_tool_submit with the token and structured output
4. Repeat until the operation returns a final result instead of another prompt

Tokens are single-use UUIDs that expire after 30 minutes. If a submission fails validation, a retry_token is provided with remaining retries.