skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
owenbush/decodie-skill245 installs

decodie-ask

Ask a question about an existing Decodie learning entry and get a deeper explanation using the entry content and live source code as context. Read-only — does not create or modify entries.

How do I install this agent skill?

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

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    This skill processes local project files and configuration to provide context for learning entries. It is vulnerable to indirect prompt injection via local rule files and could potentially expose sensitive local files if the project's metadata is maliciously crafted to point to sensitive system paths.

  • Socketpass

    No alerts

  • Snykwarn

    Risk: MEDIUM · 1 issue

What does this agent skill do?

Decodie — Ask Mode

Help the user explore and deepen their understanding of existing learning entries in the .decodie/ directory.

This mode is read-only with respect to .decodie/ data.

Setup

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.

Entry Resolution

Resolve which entry the question targets using this priority:

  1. Explicit entry ID — If the question contains an entry ID (e.g., entry-1711540000-a1b2), load it directly.
  2. Keyword match — Search active entry titles and topics in index.json. Prefer exact substring matches in titles, then topic tag matches.
  3. Current session default — If no match, default to the most recently created entry in the current session.
  4. No match — If no entries exist or match, tell the user and suggest browsing index.json.

Context Loading

Once the target entry is identified, load:

  • Entry content — Full entry from the session file: explanation, code_snippet, alternatives_considered, key_concepts.
  • Live source code — For files listed in the entry's references array, read the referenced sections to provide current-state context.
  • External documentation — URLs from external_docs for reference.

Response Instructions

  • Identify the entry — Begin by naming the entry (title and ID).
  • Answer the specific question — Use entry content and live source code as primary context.
  • Go deeper — Don't repeat the entry. Explain underlying concepts, provide examples, clarify trade-offs.
  • Suggest enrichment — If the question reveals a gap in the entry, suggest re-analysis.
  • Stay conversational — Tone of a knowledgeable colleague at a whiteboard.

Important Notes

  • After answering, the interaction is complete. The user can ask again with another question.
  • This mode only reads existing data — it does not generate entries.

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-ask">View decodie-ask on skillZs</a>