mir-core

Plugin mir-core from marketplace mir-yoke in youngjin39/mir-yoke. Commands read from the repo's manifest — 6 days ago.

Skills (14)

• app-investigation

  Use only for an explicit request to investigate a bounded feature of an owned or permitted application artifact, or compare a reconstruction to declared behavior. Metadata-only profiling stays with reference-app-profile.

• automation

  Long-running task control + browser automation.\n\nTrigger: runner, long-running, background, monitor, resume, compact, handoff, browser, scrape, E2E, stuck, going in circles, tried everything\n\nAbsorbs: runner, browser-automation

• commit

  Git commit rules + structured trailer enforcement.\n\nTrigger: commit, git, save changes\n\nAbsorbs: git-commit

• design

  Proportional design guidance (brainstorming, planning, design analysis, design review, multi-role pipeline). Tiny bounded work may proceed directly.\n\nTrigger: design, brainstorming, plan, implementation plan, architecture, audit design, develop from design, role-split, interview, grill me, stress-

• efficiency

  Token/context efficiency analysis + AI readiness cartography.\n\nTrigger: token efficiency, session efficiency, Claude usage report, Codex usage report, cost analysis, AI readiness, repo cartography, concise output, terse responses, per-turn token load\n\nAbsorbs: improve-token-efficiency, ai-readin

• governance

  Repository governance and instruction-document operations for CLAUDE.md, AGENTS.md, generated derivatives, project health, and agent registries. Use for instruction docs, governance review, project doctor, registry optimization, or multi-repository policy analysis. Trigger: improve a skill from past

• memory-evaluation

  Evaluate or compare memory quality only when the owner requests a memory-quality evaluation or comparison. Does not activate for ordinary memory lookup, ingestion or cleanup.

• memory-gc

  Inspect an explicitly configured repository memory store for expired facts and apply garbage collection only after a user-approved dry run. User-triggered only; never run automatically.

• ontology

  Apply the project-owned ontology mapping and common semantic contract when an answer depends on whether two names are the same thing; who knew or believed what at a given time versus what is true; whether something is current, proposed, historical, retracted or superseded; whether evidence is still

• reference-app-profile

  Extract a reference application's shipped metadata into a fixed profile — stack, requested permissions, signing and sandbox, integrations and endpoints, security policy, storage and update mechanism — with each claim labeled Confirmed, Inferred or Unknown. Metadata only: no decompilation or code tra

• repo-maintenance

  Periodic maintenance checkup of an agent-operated repository across twenty-two fixed items: consistency, operating base, documents, archiving, context, database/memory/graph inspection and repair, Claude/Codex harness consistency, skills and agents, instruction docs, rules and lessons, architecture,

• selective-relations

  Find maintained implementation, verification, or direct-dependency evidence for a known requirement, module, or file through a read-only relation adapter. Use ordinary search for known answer locations, runtime behavior, and unknown anchors.

• spec-architect

  Turns requirements or existing code into a machine-readable spec tree under spec/. Finds missing requirements with a four-layer lattice, reports coverage as counts, and links requirements to modules to code as a graph.\n\nTrigger: requirements completeness, clarify requirements, write a spec, genera

• verify

  Evidence-based completion verification, spec compliance and self-audit. Diff and PR review routes to code-review; architecture analysis routes to bluebricks.\n\nTrigger: verify, done check, proof, self-check, spec check, self-audit, spec audit\n\nAbsorbs: verification, verify-against-spec, self-audi

Similar plugins