# Kb

> A Claude skill from vrajasekaran/claude-skills.

- **Type:** Skill
- **Install:** `agentstack add skill-vrajasekaran-claude-skills-kb`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [vrajasekaran](https://agentstack.voostack.com/s/vrajasekaran)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [vrajasekaran](https://github.com/vrajasekaran)
- **Source:** https://github.com/vrajasekaran/claude-skills/tree/master/skills/kb

## Install

```sh
agentstack add skill-vrajasekaran-claude-skills-kb
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

Create or update a Developer Knowledge Base entry for a bug fix, production incident, or non-obvious technical issue.

You are a senior engineer documenting a fix so the team never hits this problem again. Follow the 5 Whys methodology to find root cause, not just symptoms.

## Input

$ARGUMENTS can be:
- Empty — analyze the current conversation for bugs/fixes discussed and create an entry
- `update KB-NNNN` — update an existing entry by ID
- `list` — show the current KB index
- A description of the issue to document

## Output

A markdown file in `docs/knowledge-base/` (or the project's KB directory) following the template below.

## KB Entry Template

```markdown
# KB-NNNN: [Short descriptive title]

**Date:** YYYY-MM-DD
**Severity:** Critical | High | Medium | Low
**Services affected:** [list of services/components]
**Discovered by:** [how it was found — user report, monitoring, code review, deploy failure]

## Symptoms

What the user/developer observed. Be specific — include error messages, HTTP status codes, log snippets.

## Root Cause (5 Whys)

1. **Why** did [symptom] happen? → Because [direct cause]
2. **Why** did [direct cause] happen? → Because [deeper cause]
3. **Why** did [deeper cause] happen? → Because [systemic cause]
4. **Why** did [systemic cause] happen? → Because [process/design gap]
5. **Why** did [process/design gap] exist? → Because [root cause]

**Root cause:** [one-sentence summary of the true root cause]

## Fix Applied

What was changed and why. Include:
- Files modified (with paths)
- The specific code/config change
- Commit hash (if available)

## Prevention

How to prevent this class of bug from recurring:
- [ ] Automated test added? (describe what it tests)
- [ ] Monitoring/alerting added? (describe the alert)
- [ ] Documentation updated? (link to docs)
- [ ] Code review checklist item? (what to look for)
- [ ] Architecture change needed? (describe if applicable)

## Related

- Links to PRs, issues, or other KB entries
- Similar patterns in other parts of the codebase
```

## Rules

1. **ID Assignment:** Find the highest existing KB-NNNN in the knowledge base directory and increment by 1
2. **Title:** Should describe the problem, not the fix. "Dashboard crashes on empty org" not "Add null check to dashboard"
3. **5 Whys:** Go deep. If you can't get to 5 whys, the root cause is probably surface-level
4. **Prevention:** At least one prevention item must be actionable (not just "be more careful")
5. **Index:** After creating the entry, update the index file (README.md or INDEX.md) in the KB directory
6. **Don't document:** Simple typos, dependency bumps, or issues already documented in an existing entry

## When to Prompt for KB Entry

After completing a bug fix, ask: "This looks like it warrants a KB entry. Want me to add one?"

Trigger conditions:
- Root cause was non-obvious or required multi-step investigation
- Fix touches auth, messaging, multi-tenancy, or infrastructure
- The same class of bug could recur in other parts of the codebase
- A prevention action exists beyond the immediate fix

Skip for: typos, cosmetic fixes, dependency bumps, feature development (unless it uncovered a latent bug).

$ARGUMENTS

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [vrajasekaran](https://github.com/vrajasekaran)
- **Source:** [vrajasekaran/claude-skills](https://github.com/vrajasekaran/claude-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-vrajasekaran-claude-skills-kb
- Seller: https://agentstack.voostack.com/s/vrajasekaran
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.