Agent skill · Audited from source
project-graveyard
skill · shubhamsaboo/awesome-llm-appsApache-2.0 licenceAudited 31 Oct 2026 · source b9a821973e00
Is it safe? Yes — SAFE, grade B (89/100). We read this skill's source at commit b9a821973e00 and found no filesystem, network or shell access and no hard-coded secrets. What it does: Scans the developer's machine for dead side projects, autopsies each one from its git history (died at the payments wall, killed by a newer project, finished but never shipped), surfaces their personal death patterns,…
Cite
Cite this page
ModelsAtlas. “project-graveyard (shubhamsaboo/awesome-llm-apps)”. Retrieved 31 Oct 2026. https://modelsatlas.com/skills/shubhamsaboo/project-graveyard
Trust audit
Audit 0.4.1 · source read 31 Oct 2026
SAFE means nothing in the source contradicts what the skill says it does. Grade B is shared by 16,667 of 17,602 audited skills.
What the source does
- FilesystemNone observedNo reads or writes in the source
- NetworkNone observedNo outbound calls in the source
- ShellNone observedNo commands run in the source
- DependenciesPinnedNothing floats
- SecretsNone foundNo keys in source
Five audit layers
| Layer | What we check | Result | Findings |
|---|---|---|---|
L0 Provenance | Repo owner, licence, archive status, activity | PASS | 0 |
L1 Static source | Secrets, path traversal, supply chain, hidden code | PASS | 0 |
L2 Instructions | Prompt injection, authority framing, hidden directives | PASS | 0 |
L3 Inventory | The files and declarations match the source | PASS | 0 |
L4 Sandbox | Run in isolation and watch behaviour | SKIPPED why | 0 |
Gates applied: no behavioural pass.
Findings (0)
No findings at any severity. We re-audit when the source changes.
Audit 0.4.1 · source sha b9a821973e00 · 31 Oct 2026Report an issue or request a re-scan
Provenance
Who published it, and what the verdict is pinned to
- Publishershubhamsaboo GitHub userOwner of the repository this skill is read from
- Source foldershubhamsaboo/awesome-llm-apps/agent_skills/project-graveyardThe exact folder we audited
- LicenceApache-2.0As the repository states it
- Commit pinned for this verdictb9a821973e00A new audit runs when the source changes
- Repository statusNot archivedLast push 8 days ago
- Sandboxed runNot runSkills are read, not executed; layer L4 is skipped
What it tells the agent
Full source on GitHub ↗Its description: Scans the developer's machine for dead side projects, autopsies each one from its git history (died at the payments wall, killed by a newer project, finished but never shipped), surfaces their personal death patterns,…
--- name: project-graveyard description: >- Scans the developer's machine for dead side projects, autopsies each one from its git history (died at the payments wall, killed by a newer project, finished but never shipped), surfaces their personal death patterns, and picks the corpse most worth resurrecting — then helps ship it. Use when the user mentions abandoned, unfinished, or old side projects, asks "what should I finish", wants to revive or resurrect a project, says "run the graveyard", wonders why they never finish anything, or is about to start a new project that sounds like one they already built. Runs entirely locally. license: Apache-2.0 metadata: author: "Shubham Saboo" version: "1.0.0" source: "https://github.com/Shubhamsaboo/awesome-llm-apps" --- # Project Graveyard Every developer has a folder full of dead projects. Nobody has ever gotten an autopsy report. This skill scans the machine for abandoned repos, works out why each one died from its git history, finds the user's personal death patterns, and picks the one corpse worth digging up — then helps ship it. Everything runs locally. No API, no network, nothing leaves the machine. ## When to use - The user asks about abandoned/unfinished/old side projects, or what to finish - The user wants to revive, resurrect, or "finally ship" something - The user asks why they never finish projects - The user proposes a new project — check the graveyard first (see Necromancer mode). There's a decent chance they already built half of it. ## When not to use - Cleaning up disk space or node_modules — that's `kondo`/`npkill`, not this - Archiving repos on GitHub — this works on local clones and never-pushed work - Analyzing one specific repo's history in depth — just read the git log ## Run it ```bash python3 scripts/graveyard.py ~/dev ~/projects ``` Point it at wherever projects actually live. If you don't know where that is, ask — one question beats sweeping someone's home directory uninvited. No args scans the usual suspects (~/dev, ~/projects, ~/code, ~/Desktop, ...). Useful flags: - `--days 90` — how long silent before a repo counts as dead (default 45) - `--json report.json` — full machine-readable data - `--me [email protected]` — claim commits made under other emails (repeatable); without it, projects committed via a work identity or a builder tool get skipped as "not yours" - `--include-foreign` — also include repos the user barely committed to (skipped by default: clones, forks, work checkouts are not their corpses) - `--state FILE` — remember scans and resurrections; enables relapse watch The script is read-only. It never writes inside a scanned repo. ## Reading the report The script gives you four blocks: census, the dead (with cause of death), patterns, and the top 3 by "pulse" (resurrectability score). Causes are evidence-based guesses, not verdicts — each comes with the evidence line that justifies it. If a cause looks wrong, check the evidence before repeating it. The cause taxonomy and what each one means for resurrection is in `references/causes-of-death.md` — read it before writing the report. ## The autopsy interview The script can only read git. You can ask. For corpses whose primary cause is `unknown` or `slow_fade` — the low-confidence verdicts — ask the user one question each, two or three total at most: > "`recipe-scraper` — the history just shows it drifting. Do you remember > what actually stopped you?" Blend the answers in, and label verdicts honestly in the report: **(forensic)** for what git showed, **(confirmed)** for what the user told you. Testimony beats a forensic guess — update the tombstone, not just the prose. Two questions is a conversation; five is a deposition. ## Writing the tombstone report Turn the script output into a report the user will actually feel. Format: 1. **The census.** Deaths, combined lifespan, oldest corpse. Plain numbers — they land on their own. 2. **Tombstones.** One per dead project, worst-to-best pulse. Name, lifespan, commit count, cause of death with its evidence, and a one-line epitaph. Above ~10 corpses, give full tombstones only to the 6-8 most interesting (highest pulse, most distinctive deaths) and bury the rest together in one line — "...plus 11 one-day experiments, buried in a shared plot." A wall of 23 tombstones kills the funeral. 3. **The patterns.** This is the part they'll remember. "Your projects die at day 19." "Four of six were killed by a newer project." Quote the script's numbers; add anything you can see that it can't. 4. **The resurrection.** One project. Not three. See below. Carve the resurrection pick (and only it) as an ASCII tombstone card — stone is earned, not sprayed; every other corpse stays a line in the table: ``` .------------------------. / \ | tab-sensei | | Apr — May 2026 | | 54 commits · 12 days | | | | died of deploy fear | | | | "It worked. It just | | never shipped." | ___|____________________________|___ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ``` Center the text, keep the card under 44 columns so it never wraps, epitaph last. If the graveyard is empty, no card — don't carve a stone for nobody. Epitaph rules — this is where the whole thing lives or dies: - Every epitaph must be traceable to evidence from the scan. "Died as it lived: configuring webpack" works because the config ratio was 78%. Made-up jokes about code you haven't seen don't work and the user will know. - Punch at the pattern, not the person. "It worked. It just never shipped" is fine. "You were too scared to ship" is not. - Dry beats wacky. One sentence. No puns unless they're earned. - If a project deserves respect, give it. A 26-commit repo with a finished README that never shipped is a small t
Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.
Install
Pinned to the audited commit b9a8219
npx skills add https://github.com/Shubhamsaboo/awesome-llm-apps/tree/main/agent_skills/project-graveyard
git clone https://github.com/shubhamsaboo/awesome-llm-apps git -C awesome-llm-apps checkout b9a821973e00
Commands as the repository documents them: shown, never run by us. Our verdict covers commit b9a821973e00 only.
Works in which agents?
What its own text says, not a test
| Agent | Status | More |
|---|---|---|
| Claude Code | Mentioned in its text, not tested | Skills for Claude Code · About Claude Code |
| Codex | Mentioned in its text, not tested | Skills for Codex · About Codex |
| GitHub Copilot | Mentioned in its text, not tested | Skills for GitHub Copilot · About GitHub Copilot |
| Cursor | Mentioned in its text, not tested | Skills for Cursor · About Cursor |
Audit history
Every audit, every source change
- ✓SAFE B 89 / 100
first audit · source
b9a821973e00
README badge
For maintainers · shows the latest audit
[](https://modelsatlas.com/skills/shubhamsaboo/project-graveyard/)
The badge shows our verdict and grade for this skill, and links back to this page. It changes when the audit does.
Similar skills
All coding & review skills →shubhamsaboo/awesome-llm-apps
Use when a task is too large for one model pass, needs parallel research or generation across many subtasks (like researching a dozen competitors at once), or the user asks to orchestrate multiple models, split work…
shubhamsaboo/awesome-llm-apps
Use when a task is too large for one model pass, needs parallel research or generation across many subtasks (like researching a dozen competitors at once), or the user asks to orchestrate multiple models, split work…
assafelovic/gpt-researcher
Autonomous deep research from Codex via MCP
Enoch-Nemili/skillbench
Review and harden a Model Context Protocol (MCP) server before it leaves localhost - transport and bind address, bearer or OAuth auth, Origin validation, tool input limits, file and URL access, secrets, errors and…
pytorch/pytorch
Triages GitHub issues by routing to oncall teams, applying labels, and closing questions. Use when processing new PyTorch issues or when asked to triage an issue.
pytorch/pytorch
Guide users through creating Agent Skills for Claude Code. Use when the user wants to create, write, author, or design a new Skill, or needs help with SKILL.md files, frontmatter, or skill structure.
Add this badge to your README
[](https://modelsatlas.com/skills/shubhamsaboo/project-graveyard/)
<a href="https://modelsatlas.com/skills/shubhamsaboo/project-graveyard/"><img src="https://modelsatlas.com/skills/shubhamsaboo/project-graveyard/badge.svg" alt="ModelsAtlas: SAFE B"></a>
Questions about project-graveyard
What does the project-graveyard skill do?
Is project-graveyard safe to install?
What can project-graveyard access on my machine?
Which assistants does project-graveyard work with?
How current is this page?
Sources
- github.com/shubhamsaboo/awesome-llm-apps/tree/main/agent_skills/ — the audited source at commit b9a821973e00 and every finding aboveObserved · 31 Oct 2026
Spotted something wrong? Report a correction · Methodology
Keep going in your AI
Opens with this page as the source. We add nothing else to the prompt.
- What does the project-graveyard skill do?
- Is project-graveyard safe to install?
- What can project-graveyard access on my machine?