Skip to main content

pyrunir-mcp

MCP server exposing pyrunir tools for planning-and-learning agents.

Roles

Set PYRUNIR_MCP_ROLE before launching the server. The server and invoke CLI fail closed when the role is missing, so restricted agents must be launched with an explicit role:

  • kr/ps/base: sketch-policy proof, execution, and formatting tools.
  • kr/ps/ext: module-program proof, structural termination, execution, and formatting tools.
  • kr/uns: unsolvability classifier proof and formatting tools.
  • all: every pyrunir MCP tool; use only for trusted, unrestricted local maintenance.

The server rejects missing or unknown roles at startup.

Output format

Every artifact is rendered to .psv, .md, and .json by default. Set PYRUNIR_MCP_OUTPUT_FORMAT to write a single format instead, so a consumer reads exactly one file per artifact (no adjacent duplicate renderings):

  • psv / md / json: write only that format.
  • all (or unset): write every format (the default).

(The execute manifest.json and the per-failure failures/<id>/meta.json — machine-readable run/failure metadata — are always written regardless of this setting.)

Tool Documentation

Per-tool calling arguments and normalized output structures are documented in docs/index.md. The individual tool pages cover:

  • runir.ps.base.create_empty_policy, runir.ps.base.reformat_policy, runir.ps.base.execute_policy, and runir.ps.base.prove_policy
  • runir.ps.ext.create_empty_module_program, runir.ps.ext.reformat_module_program, runir.ps.ext.reformat_module, runir.ps.ext.execute_module_program, runir.ps.ext.prove_module_program, and runir.ps.ext.prove_termination
  • runir.uns.create_empty_classifier, runir.uns.reformat_classifier, and runir.uns.prove_classifier

Start with the index for shared result conventions, especially the distinction between counterexample witness files and path trace files.

Output Contract

Proof and execution tools write a local, grouped tree under the requested output_dir: run-global alias dictionaries under dicts/, and everything for one failure local to failures/<id>/ (its meta.json, witness, and — when available — trace and successors). If that directory already contains output, the tool allocates a numbered child directory such as run-002 instead of overwriting. The selected output directory contains a .pyrunir-mcp-output reservation marker. Results include primary orchestration fields, a structured summary, and items with absolute paths to each failure's witness file and, when available, its trace. See docs/index.md for the exact shared layout (including meta.json) and the per-tool pages for argument tables.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pyrunir_mcp-0.0.17.tar.gz (137.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

pyrunir_mcp-0.0.17-py3-none-any.whl (78.2 kB view details)

Uploaded Python 3

File details

Details for the file pyrunir_mcp-0.0.17.tar.gz.

File metadata

  • Download URL: pyrunir_mcp-0.0.17.tar.gz
  • Upload date:
  • Size: 137.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for pyrunir_mcp-0.0.17.tar.gz
Algorithm Hash digest
SHA256 d6fb5a0eb8578b2a50d61b3eed2867601c738041360ba71866ad59e8fef5cfbb
MD5 c9daad4b6b9d70f2b48e4bf029d51f57
BLAKE2b-256 757d78dc49dd171550eefa3501fa74f7474e5a5172c0befa49c519ea3ced236e

See more details on using hashes here.

Provenance

The following attestation bundles were made for pyrunir_mcp-0.0.17.tar.gz:

Publisher: release.yml on planning-and-learning/runir-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file pyrunir_mcp-0.0.17-py3-none-any.whl.

File metadata

  • Download URL: pyrunir_mcp-0.0.17-py3-none-any.whl
  • Upload date:
  • Size: 78.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for pyrunir_mcp-0.0.17-py3-none-any.whl
Algorithm Hash digest
SHA256 59cfc48edae8948ae745b01756b0452f73b5394e7f410b4f7b6baa3a18bf9f4a
MD5 04cd10615e4bc2e7b0aba5d62776814e
BLAKE2b-256 be31faba7e41d8503587bae53144ce704ef342f38a92be1faf745a1f6d5a9689

See more details on using hashes here.

Provenance

The following attestation bundles were made for pyrunir_mcp-0.0.17-py3-none-any.whl:

Publisher: release.yml on planning-and-learning/runir-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page