mcp-harness-repo-maintainer

Maintains repo-local action contracts and harnesses with thin CLI and MCP adapters over core libraries.

2|Updated Aug 15, 2026
One-click install
npx skills add https://github.com/Arenukvern/shortly --skill mcp-harness-repo-maintainer-arenukvern
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mcp-harness-repo-maintainer
Source: https://github.com/Arenukvern/shortly/tree/main/.agents/skills/mcp-harness-repo-maintainer
Command: npx skills add https://github.com/Arenukvern/shortly --skill mcp-harness-repo-maintainer-arenukvern

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Repositories that expose agent-facing CLI and MCP surfaces often drift into duplicated logic, unproven tool promotions, and unverifiable adoption claims. This Skill provides governance procedures for keeping adapters thin over core libraries, proving harness readiness through cold-start benchmark loops, and promoting capabilities only with falsifiable evidence. ## Core Features & Use Cases - Action Contract Maintenance: Defines and validates steward.yaml actions, quick-safe probes, and scenario manifests with owner, risk class, effects, and redaction policy. - Cold-Start Proof Loop: Runs doctor, schema checks, action inspection, probes, and strict benchmarks to classify repos on the H0-H5 adoption maturity ladder. - CLI/MCP/Core Parity Enforcement: Ensures every MCP tool calls the same core entrypoint as its CLI twin, with archetype-specific checklists for product MCP, platform libs, CLI harness, visual sidecar, and meta-governance repos. - Use Case: When adopting Skill Steward in a new repository, use this Skill to declare a first quick-safe action, run the proof loop until a benchmark returns pass, and capture repeated friction as unknown cases before promoting typed action candidates. ## Quick Start Ask the agent to bootstrap a steward.yaml contract in this repository and run the cold-start proof loop to verify harness readiness.

Frequently Asked Questions about mcp-harness-repo-maintainer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I bootstrap a repo harness with steward.yaml?▼

Declare one quick-safe inspect-only action in steward.yaml, add a scenario manifest under steward/scenarios, then run the proof loop: steward doctor, schema checks, actions list, action inspect, probe, and benchmark with strict output. A passing benchmark summary proves H2 readiness.

How do I keep CLI and MCP adapters thin over a core library?▼

Implement each capability once in the core package with unit tests, then expose a CLI subcommand and an MCP tool that both call the same core entrypoint with identical schemas and error shapes. Divergence between the two surfaces is treated as a bug.

When should repeated agent friction become a Steward action?▼

Capture an unknown case first, then promote a typed action candidate only after owner, effects, limits, redaction, validation command, and benchmark evidence exist. Never promote diagnostics from the same run that discovered them.

What is the difference between capability-level H5 and a fully adopted repo?▼

Capability-level H5 means one named capability has repeated evidence including a held-out benchmark or future-agent repeat. A single polished proof does not make a repo fully adopted; repo maturity is a separate, broader claim.

Why does a strict benchmark return durability_blocked?▼

durability_blocked means strict benchmark inputs such as steward.yaml or scenario manifests were modified or untracked. Commit and clean declared inputs, then rerun; it is truthful blocked evidence, not H2 proof.

When should I use repo-quality-system-lifecycle instead of this Skill?▼

Use repo-quality-system-lifecycle first for general app, library, tool, or plugin stewardship baselines. Use mcp-harness-repo-maintainer only when typed actions, probes, benchmarks, or CLI/MCP parity work is in scope.