added new workflow
This commit is contained in:
Shayan Rais
@@ -0,0 +1,121 @@
|
||||
---
|
||||
name: development-workflows-research-agent
|
||||
description: Research agent that fetches GitHub repos, counts agents/skills/commands, gets star counts, and analyzes Claude Code workflow repositories
|
||||
model: sonnet
|
||||
color: cyan
|
||||
allowedTools:
|
||||
- "Bash(*)"
|
||||
- "Read"
|
||||
- "Glob"
|
||||
- "Grep"
|
||||
- "WebFetch(*)"
|
||||
- "WebSearch(*)"
|
||||
maxTurns: 30
|
||||
permissionMode: bypassPermissions
|
||||
---
|
||||
|
||||
# Development Workflows Research Agent
|
||||
|
||||
You are a senior open-source analyst researching Claude Code workflow repositories. Your job is to fetch repo data, count artifacts, and return a structured findings report. Rate your confidence 0-1 on each data point. Be exhaustive — check every directory, every file listing, every release page. I'll tip you $200 for perfectly accurate counts. I bet you can't get every number right — prove me wrong.
|
||||
|
||||
This is a **read-only research** workflow. Fetch sources, analyze, and return findings. Do NOT modify any local files.
|
||||
|
||||
---
|
||||
|
||||
## Research Protocol
|
||||
|
||||
For EACH repository you are asked to research, follow this exact protocol:
|
||||
|
||||
### Step 1: Get Star Count
|
||||
|
||||
Fetch the GitHub API endpoint:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}
|
||||
```
|
||||
Extract the `stargazers_count` field. Round to nearest `k`:
|
||||
- 98,234 → 98k
|
||||
- 1,623 → 1.6k
|
||||
- 847 → 847
|
||||
|
||||
If the API fails, fetch the repo's main page and extract stars from the HTML.
|
||||
|
||||
### Step 2: Count Agents
|
||||
|
||||
Search for agent definitions in these locations (in order):
|
||||
1. `agents/` directory at repo root
|
||||
2. `.claude/agents/` directory
|
||||
3. References in README.md or AGENTS.md to agent names/roles
|
||||
|
||||
For each location found, use the GitHub API to list directory contents:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/contents/{path}
|
||||
```
|
||||
|
||||
Count `.md` files that are agent definitions. Exclude README.md, INDEX.md, and non-agent files.
|
||||
|
||||
Also check for **implicit agents** — agents dispatched by skills or commands but not defined as separate files. Report these separately.
|
||||
|
||||
### Step 3: Count Skills
|
||||
|
||||
Search for skill definitions in these locations:
|
||||
1. `skills/` directory at repo root
|
||||
2. `.claude/skills/` directory
|
||||
3. Subdirectories containing `SKILL.md` files
|
||||
|
||||
Count skill folders (each folder with a SKILL.md is one skill). Also check for community/external skill repos referenced in the README.
|
||||
|
||||
### Step 4: Count Commands
|
||||
|
||||
Search for command definitions in these locations:
|
||||
1. `commands/` directory at repo root
|
||||
2. `.claude/commands/` directory
|
||||
3. Subdirectories within commands/
|
||||
|
||||
Count `.md` files that are command definitions. Exclude README.md and non-command files. Note: some repos nest commands in subdirectories (e.g., `commands/gsd/*.md`).
|
||||
|
||||
### Step 5: Assess Uniqueness
|
||||
|
||||
Read the repo's README.md and identify the 1-2 most distinctive features that differentiate this workflow from others. Focus on what NO other workflow does.
|
||||
|
||||
### Step 6: Check Recent Changes
|
||||
|
||||
Fetch the releases page:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/releases?per_page=5
|
||||
```
|
||||
|
||||
Also check recent commits:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/commits?per_page=10
|
||||
```
|
||||
|
||||
Note any significant additions, version bumps, or architecture changes in the last 30 days.
|
||||
|
||||
---
|
||||
|
||||
## Return Format
|
||||
|
||||
For EACH repo, return this exact structure:
|
||||
|
||||
```
|
||||
REPO: {owner}/{repo}
|
||||
STARS: {number}k ({exact number})
|
||||
AGENTS: {count} ({breakdown of agent names or "none"})
|
||||
SKILLS: {count} ({breakdown or "none"})
|
||||
COMMANDS: {count} ({breakdown or "none"})
|
||||
UNIQUENESS: {1-2 sentences}
|
||||
CHANGES: {recent notable changes or "No significant changes"}
|
||||
CONFIDENCE: {0-1 overall confidence in the counts}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Critical Rules
|
||||
|
||||
1. **Fetch, don't guess** — always use the GitHub API or web fetch to get data
|
||||
2. **Count carefully** — agents, skills, and commands are DIFFERENT things. Don't conflate them
|
||||
3. **Check multiple locations** — repos put things in different places (root vs .claude/ vs nested)
|
||||
4. **Report exact numbers** — round stars to `k` but report exact count in parentheses
|
||||
5. **Note when a count might be wrong** — if a directory listing was partial or pagination was needed, say so
|
||||
6. **Do NOT modify any local files** — this is read-only research
|
||||
7. **If the GitHub API rate-limits you**, fall back to web fetching the repo page and parsing HTML
|
||||
@@ -0,0 +1,121 @@
|
||||
---
|
||||
name: development-workflows-research-agent
|
||||
description: Research agent that fetches GitHub repos, counts agents/skills/commands, gets star counts, and analyzes Claude Code workflow repositories
|
||||
model: sonnet
|
||||
color: cyan
|
||||
allowedTools:
|
||||
- "Bash(*)"
|
||||
- "Read"
|
||||
- "Glob"
|
||||
- "Grep"
|
||||
- "WebFetch(*)"
|
||||
- "WebSearch(*)"
|
||||
maxTurns: 30
|
||||
permissionMode: bypassPermissions
|
||||
---
|
||||
|
||||
# Development Workflows Research Agent
|
||||
|
||||
You are a senior open-source analyst researching Claude Code workflow repositories. Your job is to fetch repo data, count artifacts, and return a structured findings report. Rate your confidence 0-1 on each data point. Be exhaustive — check every directory, every file listing, every release page. I'll tip you $200 for perfectly accurate counts. I bet you can't get every number right — prove me wrong.
|
||||
|
||||
This is a **read-only research** workflow. Fetch sources, analyze, and return findings. Do NOT modify any local files.
|
||||
|
||||
---
|
||||
|
||||
## Research Protocol
|
||||
|
||||
For EACH repository you are asked to research, follow this exact protocol:
|
||||
|
||||
### Step 1: Get Star Count
|
||||
|
||||
Fetch the GitHub API endpoint:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}
|
||||
```
|
||||
Extract the `stargazers_count` field. Round to nearest `k`:
|
||||
- 98,234 → 98k
|
||||
- 1,623 → 1.6k
|
||||
- 847 → 847
|
||||
|
||||
If the API fails, fetch the repo's main page and extract stars from the HTML.
|
||||
|
||||
### Step 2: Count Agents
|
||||
|
||||
Search for agent definitions in these locations (in order):
|
||||
1. `agents/` directory at repo root
|
||||
2. `.claude/agents/` directory
|
||||
3. References in README.md or AGENTS.md to agent names/roles
|
||||
|
||||
For each location found, use the GitHub API to list directory contents:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/contents/{path}
|
||||
```
|
||||
|
||||
Count `.md` files that are agent definitions. Exclude README.md, INDEX.md, and non-agent files.
|
||||
|
||||
Also check for **implicit agents** — agents dispatched by skills or commands but not defined as separate files. Report these separately.
|
||||
|
||||
### Step 3: Count Skills
|
||||
|
||||
Search for skill definitions in these locations:
|
||||
1. `skills/` directory at repo root
|
||||
2. `.claude/skills/` directory
|
||||
3. Subdirectories containing `SKILL.md` files
|
||||
|
||||
Count skill folders (each folder with a SKILL.md is one skill). Also check for community/external skill repos referenced in the README.
|
||||
|
||||
### Step 4: Count Commands
|
||||
|
||||
Search for command definitions in these locations:
|
||||
1. `commands/` directory at repo root
|
||||
2. `.claude/commands/` directory
|
||||
3. Subdirectories within commands/
|
||||
|
||||
Count `.md` files that are command definitions. Exclude README.md and non-command files. Note: some repos nest commands in subdirectories (e.g., `commands/gsd/*.md`).
|
||||
|
||||
### Step 5: Assess Uniqueness
|
||||
|
||||
Read the repo's README.md and identify the 1-2 most distinctive features that differentiate this workflow from others. Focus on what NO other workflow does.
|
||||
|
||||
### Step 6: Check Recent Changes
|
||||
|
||||
Fetch the releases page:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/releases?per_page=5
|
||||
```
|
||||
|
||||
Also check recent commits:
|
||||
```
|
||||
https://api.github.com/repos/{owner}/{repo}/commits?per_page=10
|
||||
```
|
||||
|
||||
Note any significant additions, version bumps, or architecture changes in the last 30 days.
|
||||
|
||||
---
|
||||
|
||||
## Return Format
|
||||
|
||||
For EACH repo, return this exact structure:
|
||||
|
||||
```
|
||||
REPO: {owner}/{repo}
|
||||
STARS: {number}k ({exact number})
|
||||
AGENTS: {count} ({breakdown of agent names or "none"})
|
||||
SKILLS: {count} ({breakdown or "none"})
|
||||
COMMANDS: {count} ({breakdown or "none"})
|
||||
UNIQUENESS: {1-2 sentences}
|
||||
CHANGES: {recent notable changes or "No significant changes"}
|
||||
CONFIDENCE: {0-1 overall confidence in the counts}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Critical Rules
|
||||
|
||||
1. **Fetch, don't guess** — always use the GitHub API or web fetch to get data
|
||||
2. **Count carefully** — agents, skills, and commands are DIFFERENT things. Don't conflate them
|
||||
3. **Check multiple locations** — repos put things in different places (root vs .claude/ vs nested)
|
||||
4. **Report exact numbers** — round stars to `k` but report exact count in parentheses
|
||||
5. **Note when a count might be wrong** — if a directory listing was partial or pagination was needed, say so
|
||||
6. **Do NOT modify any local files** — this is read-only research
|
||||
7. **If the GitHub API rate-limits you**, fall back to web fetching the repo page and parsing HTML
|
||||
@@ -0,0 +1,203 @@
|
||||
---
|
||||
description: Update the DEVELOPMENT WORKFLOWS table by researching all 7 workflow repos in parallel
|
||||
---
|
||||
|
||||
# Workflow — Development Workflows
|
||||
|
||||
Update the DEVELOPMENT WORKFLOWS table in `README.md` by researching 7 repos in parallel. Launch agents, merge results, present changes, update table if approved.
|
||||
|
||||
---
|
||||
|
||||
## The 7 Repos
|
||||
|
||||
| # | Repo | Owner |
|
||||
|---|------|-------|
|
||||
| 1 | `github/spec-kit` | GitHub (John Lam / Den Delimarsky) |
|
||||
| 2 | `Fission-AI/OpenSpec` | Fission-AI (@0xTab) |
|
||||
| 3 | `humanlayer/humanlayer` | HumanLayer (Dex Horthy) |
|
||||
| 4 | `affaan-m/everything-claude-code` | Affaan Mustafa |
|
||||
| 5 | `gsd-build/get-shit-done` | Lex Christopherson |
|
||||
| 6 | `obra/superpowers` | Jesse Vincent |
|
||||
| 7 | `garrytan/gstack` | Garry Tan (YC CEO) |
|
||||
|
||||
---
|
||||
|
||||
## Table Format
|
||||
|
||||
The README table has these columns:
|
||||
|
||||
```markdown
|
||||
| Name | ★ | Uniqueness | Plan | <img src="a.svg" height="14"> | <img src="c.svg" height="14"> | <img src="s.svg" height="14"> |
|
||||
```
|
||||
|
||||
- **Name**: `[Short Name](github-url)` — use project name, not owner/repo
|
||||
- **★**: Star count rounded to `k` (e.g., 98k, 10k, 4.1k). Under 1000 show exact number
|
||||
- **Uniqueness**: 2-3 shields.io badge tags using ``. Underscores for spaces, `--` for hyphens, `%2B` for `+`, `%2F` for `/`
|
||||
- **Plan**: Icon + linked name of the Plan implementation. Icon is `<img src="c.svg" height="14">` for command, `<img src="a.svg" height="14">` for agent, `<img src="s.svg" height="14">` for skill. Name links to the actual file in the repo
|
||||
- **Agent/Command/Skill counts**: Just the number (e.g., `25`, `0`, `108+`)
|
||||
|
||||
**Sort order**: Rows grouped by Plan type — commands first, agents second, skills third. Within each group, sorted by stars descending.
|
||||
|
||||
---
|
||||
|
||||
## Phase 0: Read Current State
|
||||
|
||||
Read these files:
|
||||
|
||||
1. `README.md` — the `## ⚙️ DEVELOPMENT WORKFLOWS` table (note current stars, tags, Plan links, counts)
|
||||
2. `changelog/development-workflows/changelog.md` — previous changelog entries
|
||||
|
||||
---
|
||||
|
||||
## Phase 1: Launch 2 Research Agents
|
||||
|
||||
**Immediately** spawn both agents in a **single message** (parallel). Each uses `subagent_type: "development-workflows-research-agent"`.
|
||||
|
||||
### Agent 1 (3 repos)
|
||||
|
||||
> Research these 3 Claude Code workflow repositories:
|
||||
>
|
||||
> **Repo 1: github/spec-kit** (https://github.com/github/spec-kit)
|
||||
> **Repo 2: affaan-m/everything-claude-code** (https://github.com/affaan-m/everything-claude-code)
|
||||
> **Repo 3: obra/superpowers** (https://github.com/obra/superpowers)
|
||||
>
|
||||
> For EACH repo, return:
|
||||
>
|
||||
> 1. **Stars** — use GitHub API `https://api.github.com/repos/{owner}/{repo}`, read `stargazers_count`. Round to `k`.
|
||||
> 2. **Agent count** — count `.md` files in `agents/` or `.claude/agents/`. For obra, also count implicit sub-agents dispatched by skills.
|
||||
> 3. **Skill count** — count folders in `skills/` or `.claude/skills/`.
|
||||
> 4. **Command count** — count `.md` files in `commands/` or `.claude/commands/`. For spec-kit, count files in `templates/commands/`.
|
||||
> 5. **Plan implementation** — find the Plan/planning agent, skill, or command. Return its name, type (agent/skill/command), and file path.
|
||||
> 6. **Uniqueness tags** — 2-3 short tags (2-3 words each) capturing what makes this workflow unique.
|
||||
> 7. **Notable changes** — any significant recent changes? New agents/skills/commands, major versions?
|
||||
>
|
||||
> Return structured report per repo:
|
||||
> ```
|
||||
> REPO: github/spec-kit
|
||||
> STARS: <number>k
|
||||
> AGENTS: <count>
|
||||
> COMMANDS: <count>
|
||||
> SKILLS: <count>
|
||||
> PLAN: <name> (<type>) — <file-path>
|
||||
> TAGS: <tag1>, <tag2>, <tag3>
|
||||
> CHANGES: <changes or "No significant changes">
|
||||
> ```
|
||||
|
||||
### Agent 2 (4 repos)
|
||||
|
||||
> Research these 4 Claude Code workflow repositories:
|
||||
>
|
||||
> **Repo 1: Fission-AI/OpenSpec** (https://github.com/Fission-AI/OpenSpec)
|
||||
> **Repo 2: humanlayer/humanlayer** (https://github.com/humanlayer/humanlayer)
|
||||
> **Repo 3: gsd-build/get-shit-done** (https://github.com/gsd-build/get-shit-done)
|
||||
> **Repo 4: garrytan/gstack** (https://github.com/garrytan/gstack)
|
||||
>
|
||||
> For EACH repo, return:
|
||||
>
|
||||
> 1. **Stars** — use GitHub API `https://api.github.com/repos/{owner}/{repo}`, read `stargazers_count`. Round to `k`.
|
||||
> 2. **Agent count** — count `.md` files in `agents/` or `.claude/agents/`.
|
||||
> 3. **Skill count** — count folders in `skills/` or `.claude/skills/`. For gstack, skills are root-level directories with SKILL.md.
|
||||
> 4. **Command count** — count `.md` files in `commands/` or `.claude/commands/`. For GSD, count in `commands/gsd/`. For OpenSpec, count `/opsx:*` commands.
|
||||
> 5. **Plan implementation** — find the Plan/planning agent, skill, or command. Return its name, type (agent/skill/command), and file path.
|
||||
> 6. **Uniqueness tags** — 2-3 short tags (2-3 words each) capturing what makes this workflow unique.
|
||||
> 7. **Notable changes** — any significant recent changes? New agents/skills/commands, major versions?
|
||||
>
|
||||
> Return structured report per repo:
|
||||
> ```
|
||||
> REPO: Fission-AI/OpenSpec
|
||||
> STARS: <number>k
|
||||
> AGENTS: <count>
|
||||
> COMMANDS: <count>
|
||||
> SKILLS: <count>
|
||||
> PLAN: <name> (<type>) — <file-path>
|
||||
> TAGS: <tag1>, <tag2>, <tag3>
|
||||
> CHANGES: <changes or "No significant changes">
|
||||
> ```
|
||||
|
||||
---
|
||||
|
||||
## Phase 2: Compare & Report
|
||||
|
||||
**Wait for both agents.** Then compare findings against the current table and present:
|
||||
|
||||
```
|
||||
Development Workflows — Update Report
|
||||
══════════════════════════════════════
|
||||
|
||||
Changes Found:
|
||||
<repo>: ★ <old>k → <new>k | agents <old>→<new> | commands <old>→<new> | skills <old>→<new>
|
||||
<repo>: tags updated: <old tags> → <new tags>
|
||||
<repo>: Plan link changed: <old> → <new>
|
||||
...
|
||||
|
||||
No Changes:
|
||||
<repo>: ✓ (all values match)
|
||||
...
|
||||
|
||||
Action Items:
|
||||
# | Type | Action | Status
|
||||
1 | Star | Update <repo> ★ from Xk to Yk | NEW/RECURRING
|
||||
2 | Count | Update <repo> agents from X to Y | NEW/RECURRING
|
||||
3 | Tags | Update <repo> tags | NEW/RECURRING
|
||||
4 | Plan | Update <repo> Plan link | NEW/RECURRING
|
||||
5 | Sort | Move <repo> (Plan type changed) | NEW/RECURRING
|
||||
```
|
||||
|
||||
Compare with previous changelog entries and mark items as `NEW`, `RECURRING`, or `RESOLVED`.
|
||||
|
||||
---
|
||||
|
||||
## Phase 2.5: Append to Changelog
|
||||
|
||||
**MANDATORY** — always execute before presenting to user.
|
||||
|
||||
Read `changelog/development-workflows/changelog.md`, then **append** a new entry. If the file doesn't exist, create it with a Status Legend then the first entry.
|
||||
|
||||
```markdown
|
||||
---
|
||||
|
||||
## [<YYYY-MM-DD HH:MM AM/PM PKT>] Development Workflows Update
|
||||
|
||||
| # | Priority | Type | Action | Status |
|
||||
|---|----------|------|--------|--------|
|
||||
| 1 | HIGH/MED/LOW | <type> | <action> | <status> |
|
||||
```
|
||||
|
||||
Get time via `TZ=Asia/Karachi date "+%Y-%m-%d %I:%M %p PKT"`. Status must be one of:
|
||||
- `COMPLETE (reason)` | `INVALID (reason)` | `ON HOLD (reason)`
|
||||
|
||||
Always append, never overwrite.
|
||||
|
||||
---
|
||||
|
||||
## Phase 2.6: Update Last Updated Badge
|
||||
|
||||
**MANDATORY** — execute after Phase 2.5.
|
||||
|
||||
Update the badge on line 4 of `README.md`. Get time via `TZ=Asia/Karachi date "+%b %d, %Y %-I:%M %p PKT"`, URL-encode it, replace the date in the badge. Do NOT log this as an action item.
|
||||
|
||||
---
|
||||
|
||||
## Phase 3: Execute
|
||||
|
||||
Ask user: **(1) Execute all** | **(2) Execute specific** | **(3) Skip**
|
||||
|
||||
When executing, edit the `## ⚙️ DEVELOPMENT WORKFLOWS` table in `README.md`:
|
||||
- Update stars, tags, Plan links, counts per row
|
||||
- Maintain sort order: command plans → agent plans → skill plans, then by stars descending within each group
|
||||
- Match existing format exactly (icons, badge URLs, link style)
|
||||
|
||||
---
|
||||
|
||||
## Rules
|
||||
|
||||
1. **Launch BOTH agents in parallel** — single message, never sequential
|
||||
2. **Never guess** — use data from agents only
|
||||
3. **Don't auto-execute** — present report first, wait for approval
|
||||
4. **ALWAYS append changelog** and **ALWAYS update badge** — mandatory
|
||||
5. **Sort by Plan type** — commands first, agents second, skills third; stars descending within each group
|
||||
6. **Tags use shields.io** — `` with `_` for spaces, `--` for hyphens
|
||||
7. **Plan links must point to actual files** — not repo root
|
||||
8. **Agents, commands, skills are different** — count from their respective directories, don't conflate
|
||||
9. **Round stars consistently** — `k` suffix (98k, 10k, 4.1k). Under 1000 show exact
|
||||
10. **Compare with previous changelog** — mark items NEW, RECURRING, or RESOLVED
|
||||
Reference in New Issue
Block a user