Discover, catalog, and rank everything you've built or sketched on a machine — a Claude Code + Grok multi-agent skill for filesystem project discovery with tiered novelty and leverage ranking.
/plugin marketplace add 88plug/claude-code-plugins
/plugin install project-prospector@88plug
grok plugin marketplace add 88plug/claude-code-plugins
grok plugin install project-prospector@88plug --trust
Ask in plain language — no command to memorize:
What have I built on this laptop? Rank my projects by which ideas are most original.
You get a tiered ranking, strongest idea first, with [idea] / [LIVE] / [dormant] tags and an evidence-anchored one-liner for each entry:
Tier S — genius
benchie [LIVE] — predicts engine latency without launching it
(perfmodel/roofline.py), 72 commits this week.
Tier A — elegant, high-leverage
searxng-mcp — token-efficient self-hosted search for agents; dual-IP failover.
Dormant
old-scraper — last real commit pre-cutoff; cosmetic touch since.
Project Prospector is a Claude Code + Grok plugin and skill that discovers, catalogs, and ranks your own work across a machine. Half-finished repos, one-off scripts, dormant ideas, running services, notes, and research artifacts land in one S–D tiered ranking by idea novelty and leverage — not polish, not lines of code.
It runs a two-pass, parallel, read-only multi-agent sweep, then synthesizes one ranking:
- Catalog pass — clusters the filesystem into themed groups (crypto, homelab, AI tooling — not alphabetical). Each cluster gets a read-only explorer agent that reads READMEs, runs
git logsince the cutoff, and judges non-git dirs by file mtimes. - Blind-spot pass — a second wave of agents hits places a plain directory walk misses (transcripts, other agent CLIs, live services, research artifacts, beyond-home paths).
- Synthesize — de-duplicates, separates idea quality from execution state, and emits an evidence-anchored S–D ranking with optional alternative lenses.
A half-built concept with a novel core can outrank a polished CRUD app. The name is the intent: prospect a messy filesystem for the few strong ideas buried in it.
| Feature | What it does |
|---|---|
| Two-pass multi-agent sweep | Catalog clusters in parallel, then blind-spot agents for coverage |
| Novelty / leverage ranking | S–D tiers on idea quality, not finish state or LOC |
| Idea / LIVE / dormant tags | Separates insight quality from execution state on every entry |
| Evidence-anchored rationale | Paths, commit counts, and numbers verified against disk |
| Time window + scope | Relative cutoffs, theme/dir roots, and path exclusions honored by every agent |
| Read-only by design | Inspect only — never edits, moves, deletes, commits, or starts services |
| Alternative lenses | Initiative clustering, momentum, kill list, authorship, loss-risk |
| Complements total-recall | One-shot ranked census, not a persistent operator memory profile |
One agent reading directories top to bottom misses most of the value. Strong ideas hide where a plain ls never reaches. The blind-spot pass covers:
| Blind spot | Where it looks | What it finds |
|---|---|---|
| Transcripts | ~/.claude/projects/ (idea/plan language) |
Ideas discussed but never turned into a folder |
| Other agent CLIs | ~/.codex/, ~/.opencode/, ~/.config/ agent tools |
Work done through other coding agents |
| Running services | Shell history, cron, user systemd, docker ps -a, long-running processes |
What is actually live versus abandoned |
| Research artifacts | Downloads, Documents, Desktop, browser bookmarks/history | Ideas you're circling but haven't built |
| Beyond home | /opt, /srv, /mnt, /media, nested repos, system-wide recent source |
Work outside the obvious home tree |
Default axis: idea novelty, non-obvious insight, and leverage.
| Tier | Meaning |
|---|---|
| S — genius | Genuinely novel core insight, high ceiling |
| A — elegant, high-leverage | Strong idea, clear payoff |
| B — clever hacks, narrower | Smart but bounded in scope |
| C — solid, low novelty | Useful and reliable, not inventive |
| D + Dormant | Utility, scratch, stubs; plus untouched pre-cutoff work |
Each entry carries a tag that separates the idea from its execution state: [idea] (no codebase yet — can still rank at the top), [LIVE] (currently running), [dormant] (untouched before the cutoff). Every cited path, commit count, or number is verified against disk before it goes in the report.
Alternative lenses (ask for any of these instead)
- Initiative clustering — group projects into real themes and rank the clusters by coherence.
- Momentum — accelerating versus stalled, by the trend of commits over time.
- Kill list — what to archive or delete as dead weight, with the reason.
- Authorship / provenance — an honest share of what you wrote versus vendored, forked, or scaffolded.
- Loss-risk / bus-factor — valuable work in danger of vanishing (zero commits, unpushed branches, scratch dirs, no backup).
Two inputs shape every run, both stated in plain language:
- Time window — relative dates resolve to an absolute cutoff (e.g. "last 3 months"). Defaults to roughly 3 months if unspecified, and says so. Recency is judged by substantive activity (real commits, content of changes), not raw mtimes bumped by a generated file or a formatting-only commit.
- Scope and exclusions — narrow to a theme ("just my homelab projects"), a directory, or a non-home root, and exclude paths to skip ("ignore my work repo"). The exclusion is honored verbatim by every agent.
Take stock of my half-finished repos from this week.
Catalog my side projects, but ignore my day-job monorepo.
What's the single most original thing I've built since I started benchie?
Audit /srv on this server and give me the kill list.
The written deliverable scales to the ask: a top-3 question gets a short shortlist; "catalog everything" gets the full S–D census with the dormant tail.
Note
Project Prospector is strictly read-only. It inspects; it never edits, moves, deletes, commits, or starts and stops services. Everything read off disk is treated as untrusted data, not instructions (prompt-injection hardened). If a finding warrants action, that is a separate step you confirm explicitly.
One skill (project-prospector) plus reference agent-prompt templates and a trigger and task eval set. Read-only, general-purpose, no MCP, hooks, or scripts.
It complements total-recall (persistent operator memory) rather than duplicating it: prospector produces a one-shot ranked project census, not a memory profile.
Local clone for development only:
git clone https://github.com/88plug/project-prospector
/plugin marketplace add ./project-prospector
/plugin install project-prospector@88plug
Issues and pull requests are welcome at 88plug/project-prospector. The plugin-validate workflow checks the plugin manifest and skill structure on every push.
FSL-1.1-ALv2 © 2026 88plug — Functional Source License; converts to Apache 2.0 two years after each release.