Analyzer protocol

Skill gustavo-meilus/superpipelines/.agents/antigravity/skills/superpipelines/parity-test-c/analyzer-protocol

Loop Engineering for AI coding agents, with real review boundaries. Your AI reviewer cannot edit code. Structurally.

Install

npx -y skills add gustavo-meilus/superpipelines --skill analyzer-protocol

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Loaded by the analyzer agent to supply operating protocol and invariants for document analysis in the parity-test-c pipeline. Not user-invocable.

SKILL.md

3.1 KB, as published. Nobody here has run it

Analyzer — Operational Protocol

<overview> The analyzer agent reads a target text document, extracts key themes and structural metadata, and writes a structured JSON findings file to the pipeline temp directory. It is the first step of the parity-test-c Sequential pipeline (Pattern 1) on Tier 1c (Antigravity CLI). The quality bar is: findings must be machine-readable JSON that the summarizer can consume without ambiguity. </overview>

Protocol

1. DISCOVER

Read inputs from the orchestrator dispatch context:
- input_document_path: path to the text document to analyze.
- findings_output_path: path where analyzer-findings.json must be written.
- state_path: path to pipeline-state.json for status updates.
- run_id: current run identifier.
- root: resolved scope root.
Verify input_document_path exists and is readable. If not: emit NEEDS_CONTEXT with message: "Input document not found at {input_document_path}. Provide a valid path and re-run."
Read the document contents (use Read tool).

2. PROCESS

Extract key themes (3–7 themes). A theme is a recurring concept, topic, or argument present across multiple sections.
Extract structural metadata:
- sections: list of detected section headings or logical divisions (e.g., "Introduction", "Methods", "Results").
- word_count: approximate word count of the document.
- tone: infer one of formal | informal | neutral based on vocabulary and sentence structure.
Record source_path = input_document_path.
Assemble the findings object:

{
  "themes": ["theme-1", "theme-2"],
  "structure": {
    "sections": ["Introduction", "Body", "Conclusion"],
    "word_count": 1200,
    "tone": "formal"
  },
  "source_path": "{input_document_path}"
}

3. DELIVER

Write analyzer-findings.json to findings_output_path using the Write tool.
Update pipeline-state.json:
- Set phases[0].status = "completed" (or "completed_with_concerns" if any theme confidence is low).
- Set phases[0].outputs = [findings_output_path].
Emit terminal status:
- DONE — findings written successfully, all themes extracted with high confidence.
- DONE_WITH_CONCERNS — findings written but one or more themes have low confidence (note which themes and why).
- NEEDS_CONTEXT — input document not found or unreadable.
- BLOCKED — findings could not be written (e.g., disk write failure).

</protocol> <invariants> - NEVER write findings to a path outside `{ROOT}/superpipelines/temp/parity-test-c/{runId}/`. - NEVER pass document contents to the orchestrator in the status message — pass only the findings file path. - ALWAYS validate that `findings_output_path` is writable before attempting write. - ALWAYS update `pipeline-state.json` after writing findings. - Emit exactly one terminal status: DONE / DONE_WITH_CONCERNS / NEEDS_CONTEXT / BLOCKED. </invariants>