ki-search-first

Queries project memory via ki_search before starting any task or code modification.

2|Updated Jun 8, 2026
One-click install
npx skills add https://github.com/HACK-WU/skills --skill ki-search-first-hack-wu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ki-search-first
Source: https://github.com/HACK-WU/skills/tree/main/skills/ki-search-first
Command: npx skills add https://github.com/HACK-WU/skills --skill ki-search-first-hack-wu

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Engineers repeatedly re-investigate problems that were already solved and documented in project memory, and code changes often break cross-module contracts because hidden dependencies were never checked. This Skill enforces a memory-first workflow so prior knowledge is consulted before any task begins. ## Core Features & Use Cases - Memory-First Retrieval: Runs ki_search against project memory as the mandatory first step for any task or question, using natural-language problem descriptions. - Graceful Degradation: Falls back to the expert-lookup and solution-lookup reuse path (expert-solution-workflow) when memory has no sufficient answer. - Strong-Association Checks: When code is located or modified, queries strong-association relationships of the changed module to identify affected consumers and contract partners. - Use Case: Before fixing an order-timeout bug, the agent searches project memory, finds a prior postmortem pointing to the exact handler, and after editing the data source module, checks which consumer modules must be updated in tandem. ## Quick Start Before working on my task, search the project memory with ki_search for anything related to this problem and tell me what you find.

Frequently Asked Questions about ki-search-first

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

FAQPage Schema
How do I search project memory before starting a coding task?

Invoke ki_search with a natural-language description of the problem, such as "how is order timeout handled", rather than single keywords. Review the returned memory entries for code locations, conclusions, and known pitfalls before reading source code.

What should I do when project memory search returns no results?

Fall back to the expert-solution-workflow path: classify the problem type, consult the expert-lookup skill for business module experts or solution-lookup for technical solutions, then inspect the codebase directly. After solving, persist valuable conclusions back into project memory.

Why check strong associations when modifying code?

Strong associations reveal asymmetric contracts where changing a data source requires updating its consumers, which are not obvious from reading the code alone. Querying them before or after edits prevents runtime failures caused by missed cross-module dependencies.

When can the project memory search be skipped?

Skip it only when the user explicitly says not to check memory, during casual non-task conversation, or for trivial one-off edits like fixing a single word that clearly has no relation to project knowledge.

What if project memory conflicts with the actual source code?

Treat the source code as the source of truth because memory entries can become outdated. The memory still serves as a starting point for locating relevant code, but conclusions must be verified against the current implementation.