| name | b00t-gh-issues |
| description | GitHub issue triage and integration review agent for the b00ty-verse.
Dogfoods b00t's grok RAG subsystem to research an issue topic, then
posts a structured integration review on the GH issue. Evaluates
external tools/frameworks/libraries for potential integration into
the b00t ecosystem. Output is a machine-parseable review comment
covering: datum creation, integration level, overlap analysis,
capability gaps, and recommended next actions.
|
| version | 1.0.0 |
| tags | ["github","issues","triage","grok","research","integration-review","b00ty-verse"] |
| applies_to | ["GitHub issue triage","b00t backlog grooming","integration analysis","explorer reviews"] |
| output_types | [".md"] |
| allowed-tools | Read, Write, Bash, Grep, Glob, WebFetch, Task, mcp__b00t-mcp__b00t_grok_ask, mcp__b00t-mcp__b00t_grok_learn, mcp__b00t-mcp__b00t_grok_digest |
What This Skill Does
Reviews GitHub issues in elasticdotventures/_b00t_ against the b00ty-verse
(b00t ecosystem). Uses grok RAG to research the issue topic, analyzes how the
proposed capability fits into existing b00t features, and posts a structured
integration review comment back on the issue.
Each review MUST address all 5 mandatory sections defined below.
When It Activates
- "review this issue" or "triage the backlog"
- "analyze issue #N" or "how does this fit into b00t"
- "grok this issue" or "research this GH issue"
- "b00t-gh-issues" or "run issue review"
just gh-issue-review #<N>
- After reopening previously closed exploratory issues
Review Workflow
1. FETCH issue → gh issue view #N --json number,title,body,labels,comments
2. EXTRACT → URLs, tool names, framework names, key concepts from issue body+comments
3. GROK-RESEARCH → b00t grok ask for each key concept (existing knowledge check)
4. GROK-INGEST → b00t grok learn any new URLs mentioned in the issue
5. B00T-LEARN → b00t learn <concept> for each concept (check datum coverage)
6. OVERLAP-SCAN → grep/grok over codebase for existing implementations of similar ideas
7. COMPOSE → structured review per the output format below
8. POST → gh issue comment #N --body-file review.md
Step Detail
1. FETCH
gh issue view "$ISSUE_NUM" --repo elasticdotventures/_b00t_ \
--json number,title,body,labels,comments --jq '.' > /tmp/b00t-issue-$ISSUE_NUM.json
2. EXTRACT
Parse the issue for:
- URLs: any
https:// links to repos, docs, papers, tools
- Tool names: identifiable software projects, frameworks, libraries
- Concepts: key technical ideas the issue is exploring
3. GROK-RESEARCH
For each key concept, query grok's RAG:
b00t grok ask "<concept description>" -t "$SLUG" --limit 3
Check grok's existing knowledge about the topic. If grok returns results, note
what b00t already knows.
4. GROK-INGEST
If the issue mentions external URLs not yet in b00t's knowledge base:
b00t grok learn --source-url "$URL" -t "$SLUG"
Or use WebFetch to read the URL, then digest:
b00t grok digest -t "$SLUG" "$CONTENT"
5. B00T-LEARN
Check existing datum coverage for each concept:
b00t learn "$CONCEPT" --concise 2>/dev/null || echo "NO_DATUM_FOUND"
If no datum exists, this flags a knowledge gap that may need research-soul.
6. OVERLAP-SCAN
Search the codebase for existing implementations that overlap:
- Grep for the tool name, concept, or pattern
- Check
_b00t_/ for related datums
- Check
justfile for related recipes
- Check
b00t-cli/src/ for related commands
- Note any existing MCP servers, skills, or agents that overlap
7. COMPOSE
Synthesize findings into the structured output format. Be laconic, precise,
and opinionated. This is not a summary — it's a recommendation with evidence.
8. POST
gh issue comment "$ISSUE_NUM" --repo elasticdotventures/_b00t_ --body-file /tmp/b00t-review-$ISSUE_NUM.md
Output Format (STRICT)
The review comment MUST follow this exact structure:
## b00ty-verse Integration Review
**Reviewer**: {b00t-agent-name} | **Model**: {model-id} | **Tier**: {sm0l|ch0nky|frontier}
**Date**: {YYYY-MM-DD}
### Summary
{1-paragraph synthesis of what this issue proposes and how it fits into b00t}
### 1. Datums Being Created
- `datum-name.skill.toml` — {what it configures, why needed}
- `datum-name.mcp.toml` — {what MCP service it registers}
- (or "None — no new datums required")
### 2. Integration Level
**Verdict**: DEPENDENCY | MINE-AND-DISTILL | FORK-PATCH-INTEGRATE | DECLINE
{1-2 sentences justifying the verdict. If DECLINE, explain why.}
### 3. Existing Feature Overlap & Complement
| Existing b00t Feature | Relationship | Detail |
|-----------------------|--------------|--------|
| `b00t grok` | Overlap / Complement | {explanation} |
| `firecrawl` skill | Overlap / Complement | {explanation} |
| ... | ... | ... |
### 4. Capability Gaps (Ranked by Utility)
| Rank | Gap | Current State | Utility (1-10) |
|------|-----|---------------|----------------|
| 1 | {gap description} | {how b00t handles it now} | N |
| 2 | ... | ... | N |
### 5. Recommended Next Actions
1. **{action verb}**: {concrete, verifiable step}
2. **{action verb}**: {concrete, verifiable step}
3. **{action verb}**: {concrete, verifiable step}
---
Reviewed by: {agent-name} ({model}) | Tier: {tier} | b00t-gh-issues v1.0.0
Output Rules
- Be laconic — no platitudes, no praise, no apologies
- Ranked lists must be in descending priority order
- "MINE-AND-DISTILL" means: the content/tool has useful ideas but the code itself won't be integrated
- "FORK-PATCH-INTEGRATE" means: the tool will be forked, potentially patched, and integrated as a b00t dependency or submodule
- "DEPENDENCY" means: b00t will add it as a direct dependency (crate, npm package, etc.)
- "DECLINE" means: the proposal does not fit the b00ty-verse (explain why)
- Capability gap utility is: 1-3 = niche, 4-6 = useful, 7-9 = important, 10 = critical
- If recommending datums, specify exact filenames and types
Reference: b00ty-verse Components
The b00ty-verse includes:
| Component | Purpose |
|---|
b00t-cli | Main CLI — commands, learn, task, grok, hive |
b00t-c0re-lib | Shared Rust library — grok client, dual grok, irontology bridge |
b00t-c0re-a2a | Agent-to-agent communication |
b00t-mcp | MCP server exposing b00t tools |
b00t-grok-py | Python grok server (Qdrant + Ollama + RAG-Anything) |
b00t-grok | Rust PyO3 bindings for grok |
b00t-j0b-py | Web crawler and job processor |
ledg3rr | b00t's Rust interface library (process lifecycle, governance) |
irontology-mcp | Semantic graph/RAG backend (vendor submodule) |
_b00t_/ | Datum directory — TOML configurations for all tools/services |
skills/ | Agent skills (SKILL.md + YAML frontmatter) |
AGENTS/ | Role supplements (--role=*.md) |
Reference: b00t grok Commands
b00t grok ask "<query>" -t <topic> --limit 5
b00t grok learn -s <file|URL> "<content>" -t <topic>
b00t grok digest -t <topic> "<text>"
b00t grok assimilate -t <topic> --source-url <URL> --enhanced
b00t grok status
Justfile Integration
just gh-issue-review 424
just gh-issue-review --backlog
Example Usage
Single Issue Review
User: review issue #424 (integrate sonar)
Agent:
1. FETCH → issue #424 body: "useful debugging tool for b00t, https://github.com/RasKrebs/sonar"
2. GROK-ASK → "sonar debugging tool" → no existing knowledge
3. WebFetch → https://github.com/RasKrebs/sonar → README content
4. GROK-LEARN → ingest sonar README content
5. B00T-LEARN → "sonar" → NO_DATUM_FOUND
6. OVERLAP-SCAN → grep "sonar\|debug\|tracing" in codebase → existing tokio-console, tracing
7. COMPOSE → structured review
8. POST → gh issue comment #424
Backlog Sweep
User: run b00t-gh-issues against all reopened exploration issues
Agent:
1. Lists all open issues via gh issue list --repo elasticdotventures/_b00t_
2. For each issue without a b00t-gh-issues review comment, runs the review workflow
3. Posts individual reviews as issue comments
Version
1.0.0 — Initial b00t-gh-issues skill with grok-dogfooding integration review workflow