Project Status

Project Status

Assembles a complete picture from two sources: the memory dossier (what we know) + live data (what is happening now).

---

Step 1: Identify the project

Extract the project from the user's message. Input types:

| Input | How to identify | |-------|----------------| | GitHub URL | Extract repo name from URL | | Project name | Direct match against memory-config.json keywords | | Directory path | Last path component | | Context ("that project with the landing page") | Search dossiers by description |

Load keyword mappings:

Read ~/.claude/memory/memory-config.json

If the project cannot be identified, ask the user.

---

Step 2: Load dossier

Read ~/.claude/memory/projects/{project-name}.md

If dossier not found, warn and work with live data only.

---

Step 3: Gather live data

Run in parallel (whatever is available for the project):

3a. Git (if local repository exists)

git -C {project_path} log --oneline -10
git -C {project_path} status --short
git -C {project_path} log -1 --format="%cr" # time since last commit

3b. GitHub (if remote repo exists)

gh repo view {owner/repo} --json updatedAt,pushedAt --jq '.pushedAt'
gh issue list -R {owner/repo} --limit 5 --json number,title,labels
gh pr list -R {owner/repo} --limit 5 --json number,title,state

3c. Deploy status (if applicable)

Check last deploy if the dossier mentions a deployment platform.

---

Step 4: Generate report

Output format, compact and structured:

# {Project Name} - Status

## Pulse
- Last commit: {when}
- Last push: {when}
- Uncommitted: {yes/no, N files}

## Recent Changes (git log -10)
{commit list}

## Open Issues
### From dossier (unresolved problems):
- {problem}

### GitHub Issues:
- #{N} {title}

### Pull Requests:
- #{N} {title} ({state})

## Next Steps (from dossier)
1. {task}
2. {task}

## Since Last Session
Last session: {date}
What was done: {from dossier}

---

Rules

Speed over completeness

• Do NOT clone the repository. If no local copy exists, use GitHub API via gh
• Do NOT read project files unless the question requires it
• Dossier + git log + issues = enough for 90% of queries

Specific questions

The user may ask specifics:

• "when was the last scrape?" - search git log for commits with scrape/parse/crawl
• "what broke?" - search issues with bug label, or git log for fix commits

Update dossier if new data found

If live data reveals something not in the dossier (new commits, closed issues), suggest updating via /session-save.

Source & license

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

• Author: vasilievyakov
• Source: vasilievyakov/miracle-infrastructure
• License: MIT

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