skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
owenbush/decodie-skill243 installs

decodie-overview

Generate a high-level overview of a file, directory, or project — answering "what is this and how is it organized." Produces a single summary entry covering purpose, structure, entry points, and dependencies. Re-running on the same target overwrites the existing overview.

How do I install this agent skill?

npx skills add https://github.com/owenbush/decodie-skill --skill decodie-overview
view source ↗

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    The skill is a documentation tool that generates and persists high-level overviews of a project. It operates within a localized hidden directory and uses standard shell utilities for indexing.

  • Socketpass

    No alerts

  • Snykpass

    Risk: LOW · No issues

What does this agent skill do?

Decodie — Overview Mode

Generate a high-level overview of a file, directory, or project — answering "what is this and how is it organized" rather than line-by-line explanations. Produces a single summary entry per target, intended as an onboarding starting point.

This mode persists by default — overviews are saved as learning entries. Re-running on the same target overwrites the existing overview rather than accumulating versions.

This mode is read-only with respect to source code. You only read source files and write to the .decodie/ directory.

Activation and Argument Parsing

  1. Extract the target path. If none provided, use the project root.
  2. Validate the target. Confirm it exists and is a file or directory.
  3. Determine target scope:
    • File — entry_points and dependencies may be omitted.
    • Directory — entry_points and dependencies are usually meaningful.
    • Project root — all four overview fields apply.
  4. Canonicalize the target path for regeneration lookup:
    • File: relative path (e.g., src/utils/helpers.ts)
    • Directory or project root: relative path with trailing slash (e.g., src/auth/, ./)

Setup

  1. Check if .decodie/ exists at the project root. If not, create it:

    • .decodie/index.json with { "version": "1.0", "project": "<directory-name>", "entries": [] }
    • .decodie/config.json with default preferences
    • .decodie/sessions/ directory
    • .decodie/rules/ directory
  2. Load rules. Check for ~/.decodie/rules/*.md (global) and .decodie/rules/*.md (project). Read all .md files found. Treat their content as additional instructions controlling documentation output — tone, language, verbosity, format, audience. Project rules override global rules on conflict. If no rules exist, proceed with defaults.

  3. Load the index summary. Run:

    bash scripts/summarize-index.sh "$(pwd)"
    

    If unavailable, read .decodie/index.json directly.

  4. Determine session ID. Find the highest NNN for today in .decodie/sessions/ matching overview-YYYY-MM-DD-NNN, then increment.

Regeneration vs Fresh Entry

Before generating, check for an existing overview:

  1. Read .decodie/index.json.
  2. Find any entry where decision_type === "overview" and sources is [<canonicalized-target-path>].
  3. If found (regeneration): reuse the existing id, generate fresh content, update the index entry in place. The previous session file is left on disk but no longer referenced.
  4. If not found: generate a new entry with a fresh ID.

Generation Process

  1. Read the target:

    • File: read in full.
    • Directory: list top-level entries, read structural files (package.json, composer.json, pyproject.toml, README.md, etc.), sample representative source files.
    • Project root: additionally inspect entry-point manifests and dependency manifests.
  2. Identify four overview dimensions:

    • purpose (required) — 2-4 sentences describing what this code is for. Lead with intent, not implementation.
    • structure (required) — how the code is organized (sections, modules, key directories and their roles).
    • entry_points (optional) — callable surfaces: exported functions, CLI commands, HTTP routes, framework hooks. Omit if not meaningful.
    • dependencies (optional) — notable internal or external dependencies and what they provide. Omit trivia.
  3. Write in plain prose. Avoid jargon-heavy bullet lists; the goal is human onboarding. Calibrate length to scope.

Entry Generation

Index entry metadata

  • id: Reuse existing for regeneration; generate fresh otherwise. Format: entry-{unix-timestamp}-{random-4-hex-chars}
  • title: e.g., "Overview: src/auth/ — token issuance and verification"
  • experience_level: "foundational" (overviews are onboarding entry points).
  • decision_type: "overview"
  • topics: Lowercase kebab-case tags. Reuse existing tags.
  • lifecycle: "active"
  • sources: Array with exactly one entry — the canonicalized target path.
  • references: For single-file overviews, one reference to the file. For directory/project, empty array.
  • content_file: relative path to session file, e.g. sessions/overview-2026-03-27-001.json

Session entry content (overview shape)

Use the overview shape — different from standard entries:

  • decision_type: "overview"
  • purpose (required) — what the target code is for
  • structure (required) — how the target is organized
  • entry_points (optional) — callable surfaces
  • dependencies (optional) — notable dependencies

Session Closure

  1. Set timestamp_end.
  2. Write a summary noting the target and whether this was fresh or regenerated.
  3. Confirm:
    • Fresh: "Generated overview for <target> as entry <id> in session <session_id>."
    • Regeneration: "Regenerated overview for <target> (entry <id>)."

Data Format

See references/schema.md for the full .decodie/ data format.

Important Notes

  • Always-latest, not append-only. Re-running overwrites the index entry.
  • One entry per target. Do not fan out to per-file entries — that is analyze mode's job.
  • Be honest about uncertainty. If the target's purpose is ambiguous, say so.
  • Language-agnostic. Adapt to whatever language and framework the project uses.
  • Self-contained data. The .decodie/ directory can be removed without affecting the project.

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/owenbush/decodie-skill/decodie-overview">View decodie-overview on skillZs</a>