skills.rest
pavelvdo
pavelvdoCOMMUNITY
Explore all skills by this author →
GitHub

openspec-knowledge-add

Extracts verified knowledge base facts from standalone markdown reports into an indexed KB.

Updated Jun 26, 2026
Search Tags:#documentation#knowledge base#deduplication#openspec#fact extraction#1c#markdown reports
One-click install
npx skills add https://github.com/pavelvdo/universal-xml-exchange2 --skill openspec-knowledge-add-pavelvdo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: openspec-knowledge-add
Source: https://github.com/pavelvdo/universal-xml-exchange2/tree/main/.cursor/skills/openspec-knowledge-add
Command: npx skills add https://github.com/pavelvdo/universal-xml-exchange2 --skill openspec-knowledge-add-pavelvdo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Verified technical facts discovered during investigations often stay buried in one-off reports and are lost. This Skill turns standalone analytical reports and markdown notes into durable, indexed knowledge base entries with validated anchors, deduplication, and TTL policies. ## Core Features & Use Cases - Verified KB Extraction: Pulls up to 5 knowledge candidates per session from canonical report types (exploration, trace-analysis, architecture, and similar), validating anchors against the current source tree before saving. - Taxonomy and Dedup Enforcement: Assigns domain/subdomain from a taxonomy file, rejects duplicates of existing KB entries, and supports superseding outdated facts. - Confirmation-Gated Saving: Shows per-candidate preview cards and requires explicit user confirmation before writing KB files, rebuilding the index, and bundling unstable sources. - Use Case: After an investigation produces temp/reports/exploration-foo-2026-04-28.md, run the command to distill its verified findings into permanent KB entries instead of letting the report rot in a temp folder. ## Quick Start Ask the assistant to run /opsx:knowledge-add with the path to your markdown report, for example to extract verified facts from temp/reports/exploration-foo-2026-04-28.md into the knowledge base.

Frequently Asked Questions about openspec-knowledge-add

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

FAQPage Schema
How do I add knowledge base entries from a markdown report?▼

Run /opsx:knowledge-add with one or more paths to markdown reports, such as files in temp/reports/. The command extracts up to 5 verified candidates, shows preview cards, and saves only after your confirmation.

What report types can be used as knowledge sources?▼

Canonical sources are markdown files matching masks like exploration-*, trace-analysis-*, resolved-contract-*, architecture-*, deep-analysis-*, and design-review-*. Other markdown files work only if they contain verified facts with anchors, and code files like .bsl or .xml are rejected.

Can I extract knowledge directly from BSL or XML source code?▼

No, the command refuses .bsl, .xml, .json, and trace files as inputs. You must first produce an analytical report through an exploration workflow, then pass that report to knowledge-add.

Why was my knowledge candidate skipped or deferred?▼

Candidates are skipped for stale or unverified anchors, duplication of an existing KB entry, or taxonomy mismatch. They are deferred when the Reuse Value Test fails, meaning no concrete future reuse scenario could be justified.

What happens if I run the command without any file paths?▼

It applies a Session Context Fallback, considering only report paths already visible in the current session. If no candidates are found, it prints an error with usage examples and writes nothing.

When should I use knowledge-add instead of archive-based extraction?▼

Use knowledge-add for standalone sources outside a change archive, such as temp reports or manual notes. Use the archive flow or knowledge-audit --from-archive when extracting knowledge from an active or archived change request.

Related Skills

Explore More Software Engineering Skills→
damoke012damoke012

research

Investigates questions against primary sources and writes cited findings to Markdown files.

Community
Basic
neuron-oneneuron-one
25

graphiti

Store and query time-stamped facts in a temporal knowledge graph.

Community
Advanced
jacob-balslevjacob-balslev
1

taxonomy-design

Design controlled classification taxonomies with category trees and facet governance.

Community
Intermediate
ai-analyst-labai-analyst-lab
21

archive-analysis

Archive completed analyses into the knowledge archive during run-analysis.

Official
Basic