github-starred/awesome-llm-apps
2
0
Fork 0
mirror of https://github.com/Shubhamsaboo/awesome-llm-apps.git synced 2026-07-18 10:48:20 -05:00
Code Issues 136 Packages Projects Releases Wiki Activity

[PR #934] [MERGED] Agent Skills: remove 19 prompt-dump skills, ship project-graveyard #14082

New Issue
Closed
opened 2026-07-11 19:31:47 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-07-11 19:31:47 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/Shubhamsaboo/awesome-llm-apps/pull/934
Author: @Shubhamsaboo
Created: 7/8/2026
Status: ✅ Merged
Merged: 7/8/2026
Merged by: @Shubhamsaboo

Base: main ← Head: agent-skills-v1

📝 Commits (2)

  • 26f859f chore: remove 19 basic prompt-dump skills
  • e940f0d feat: add project-graveyard agent skill

📊 Changes

48 files changed (+1371 additions, -5924 deletions)

View changed files

➕ .github/workflows/skill-evals.yml (+34 -0)
📝 README.md (+3 -28)
📝 awesome_agent_skills/README.md (+34 -132)
➖ awesome_agent_skills/academic-researcher/SKILL.md (+0 -269)
➖ awesome_agent_skills/code-reviewer/AGENTS.md (+0 -292)
➖ awesome_agent_skills/code-reviewer/SKILL.md (+0 -135)
➖ awesome_agent_skills/code-reviewer/rules/correctness-error-handling.md (+0 -279)
➖ awesome_agent_skills/code-reviewer/rules/maintainability-naming.md (+0 -149)
➖ awesome_agent_skills/code-reviewer/rules/maintainability-type-hints.md (+0 -93)
➖ awesome_agent_skills/code-reviewer/rules/performance-n-plus-one.md (+0 -250)
➖ awesome_agent_skills/code-reviewer/rules/security-sql-injection.md (+0 -142)
➖ awesome_agent_skills/code-reviewer/rules/security-xss-prevention.md (+0 -225)
➖ awesome_agent_skills/content-creator/SKILL.md (+0 -237)
➖ awesome_agent_skills/data-analyst/SKILL.md (+0 -57)
➖ awesome_agent_skills/debugger/SKILL.md (+0 -230)
➖ awesome_agent_skills/decision-helper/SKILL.md (+0 -92)
➖ awesome_agent_skills/deep-research/SKILL.md (+0 -192)
➖ awesome_agent_skills/editor/SKILL.md (+0 -212)
➖ awesome_agent_skills/email-drafter/SKILL.md (+0 -83)
➕ awesome_agent_skills/evals/README.md (+41 -0)

...and 28 more files

📄 Description

What this does

Two commits, one story: the agent-skills section stops hosting filler and ships its first skill built to a real quality bar.

Commit 1 — remove 19 basic prompt-dump skills. Single-file persona skills (code-reviewer, email-drafter, content-creator, ...) were advice the model already gives without them. Deleted, −5,674 lines.

Commit 2 — add project-graveyard. Scans your machine for dead side projects, autopsies why each one died from its git history, shows your personal death patterns, and resurrects the one with a pulse:

  • 7 evidence-based causes of death — deploy fear (finished, never shipped), payments/auth wall, boilerplate wall, scope explosion, rewrite spiral, slow fade, and shiny object — cross-repo timeline analysis that names which newer project killed which older one
  • Agent layer on top of the scanner: autopsy interview (forensic vs confirmed verdicts), world-check (what got easier since it died), ≤7-step resurrection plan the agent starts executing, relapse watch on every resurrection it prescribes, necromancer mode ("you already built 60% of this")
  • Local and private by design: Python stdlib, zero network calls, read-only on every repo, sees never-pushed projects precisely because it doesn't use the GitHub API; --redact for shareable reports
  • Install (after merge): npx skills add https://github.com/Shubhamsaboo/awesome-llm-apps/tree/main/awesome_agent_skills/project-graveyard — installs into 70+ agents (Claude Code, Codex, Cursor, Copilot, Antigravity, OpenClaw, Hermes)

Quality gates (all passing)

  • 16/16 executable eval — awesome_agent_skills/evals/project-graveyard/test_graveyard.py builds a synthetic graveyard and asserts every classifier, kill-chain naming, finished-vs-dead separation, redaction, state/relapse, and JSON shape. Doubles as the pre-install trust check (~10s, no deps).
  • Strict SKILL.md lint, clean security scan, verified on a real machine (found a genuine deploy-fear corpse on first run).
  • The installable folder contains only runtime files — evals stay repo-side.

Also in here

  • self-improving-agent-skills drops its toy example files; its examples endpoint now lists sibling skills from this repo, so the app demos on real skills.
  • Skills READMEs updated: quality bar, one-command install, verified 2026 agent matrix.

🤖 Generated with Claude Code

Post-review fix (pushed as amended commit)

A review pass caught that the eval was 15/16 in a clean Linux checkout — the "kill-chain names the killer" check was filesystem-order-dependent: two fixture repos both qualified as the killer, and the autopsy took the first one in os.walk order (APFS and ext4 enumerate differently). Fixed properly, not by adjusting the test:

  • Nearest-killer selection: when several newer projects qualify, the closest in time gets the blame — deterministic and better semantics (name breaks exact ties)
  • Repo discovery output is now sorted, so nothing downstream depends on walk order
  • Verified under adversarial (reversed) discovery order + 3 consecutive clean eval runs: 16/16, deterministically

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/Shubhamsaboo/awesome-llm-apps/pull/934 **Author:** [@Shubhamsaboo](https://github.com/Shubhamsaboo) **Created:** 7/8/2026 **Status:** ✅ Merged **Merged:** 7/8/2026 **Merged by:** [@Shubhamsaboo](https://github.com/Shubhamsaboo) **Base:** `main` ← **Head:** `agent-skills-v1` --- ### 📝 Commits (2) - [`26f859f`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/26f859f44bca6ba7c5ff0f1ff610f4f150b7ed02) chore: remove 19 basic prompt-dump skills - [`e940f0d`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/e940f0ddc6f58a443733529d3cda327abc14d9e0) feat: add project-graveyard agent skill ### 📊 Changes **48 files changed** (+1371 additions, -5924 deletions) <details> <summary>View changed files</summary> ➕ `.github/workflows/skill-evals.yml` (+34 -0) 📝 `README.md` (+3 -28) 📝 `awesome_agent_skills/README.md` (+34 -132) ➖ `awesome_agent_skills/academic-researcher/SKILL.md` (+0 -269) ➖ `awesome_agent_skills/code-reviewer/AGENTS.md` (+0 -292) ➖ `awesome_agent_skills/code-reviewer/SKILL.md` (+0 -135) ➖ `awesome_agent_skills/code-reviewer/rules/correctness-error-handling.md` (+0 -279) ➖ `awesome_agent_skills/code-reviewer/rules/maintainability-naming.md` (+0 -149) ➖ `awesome_agent_skills/code-reviewer/rules/maintainability-type-hints.md` (+0 -93) ➖ `awesome_agent_skills/code-reviewer/rules/performance-n-plus-one.md` (+0 -250) ➖ `awesome_agent_skills/code-reviewer/rules/security-sql-injection.md` (+0 -142) ➖ `awesome_agent_skills/code-reviewer/rules/security-xss-prevention.md` (+0 -225) ➖ `awesome_agent_skills/content-creator/SKILL.md` (+0 -237) ➖ `awesome_agent_skills/data-analyst/SKILL.md` (+0 -57) ➖ `awesome_agent_skills/debugger/SKILL.md` (+0 -230) ➖ `awesome_agent_skills/decision-helper/SKILL.md` (+0 -92) ➖ `awesome_agent_skills/deep-research/SKILL.md` (+0 -192) ➖ `awesome_agent_skills/editor/SKILL.md` (+0 -212) ➖ `awesome_agent_skills/email-drafter/SKILL.md` (+0 -83) ➕ `awesome_agent_skills/evals/README.md` (+41 -0) _...and 28 more files_ </details> ### 📄 Description ## What this does Two commits, one story: the agent-skills section stops hosting filler and ships its first skill built to a real quality bar. **Commit 1 — remove 19 basic prompt-dump skills.** Single-file persona skills (code-reviewer, email-drafter, content-creator, ...) were advice the model already gives without them. Deleted, −5,674 lines. **Commit 2 — add `project-graveyard`.** Scans your machine for dead side projects, autopsies why each one died from its git history, shows your personal death patterns, and resurrects the one with a pulse: - **7 evidence-based causes of death** — deploy fear (finished, never shipped), payments/auth wall, boilerplate wall, scope explosion, rewrite spiral, slow fade, and *shiny object* — cross-repo timeline analysis that names which newer project killed which older one - **Agent layer on top of the scanner**: autopsy interview (forensic vs confirmed verdicts), world-check (what got easier since it died), ≤7-step resurrection plan the agent starts executing, relapse watch on every resurrection it prescribes, necromancer mode ("you already built 60% of this") - **Local and private by design**: Python stdlib, zero network calls, read-only on every repo, sees never-pushed projects precisely because it doesn't use the GitHub API; `--redact` for shareable reports - **Install** (after merge): `npx skills add https://github.com/Shubhamsaboo/awesome-llm-apps/tree/main/awesome_agent_skills/project-graveyard` — installs into 70+ agents (Claude Code, Codex, Cursor, Copilot, Antigravity, OpenClaw, Hermes) ## Quality gates (all passing) - **16/16 executable eval** — `awesome_agent_skills/evals/project-graveyard/test_graveyard.py` builds a synthetic graveyard and asserts every classifier, kill-chain naming, finished-vs-dead separation, redaction, state/relapse, and JSON shape. Doubles as the pre-install trust check (~10s, no deps). - Strict SKILL.md lint, clean security scan, verified on a real machine (found a genuine deploy-fear corpse on first run). - The installable folder contains only runtime files — evals stay repo-side. ## Also in here - `self-improving-agent-skills` drops its toy example files; its examples endpoint now lists sibling skills from this repo, so the app demos on real skills. - Skills READMEs updated: quality bar, one-command install, verified 2026 agent matrix. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- ## Post-review fix (pushed as amended commit) A review pass caught that the eval was **15/16 in a clean Linux checkout** — the "kill-chain names the killer" check was filesystem-order-dependent: two fixture repos both qualified as the killer, and the autopsy took the first one in `os.walk` order (APFS and ext4 enumerate differently). Fixed properly, not by adjusting the test: - **Nearest-killer selection**: when several newer projects qualify, the closest in time gets the blame — deterministic *and* better semantics (name breaks exact ties) - Repo discovery output is now **sorted**, so nothing downstream depends on walk order - Verified under adversarial (reversed) discovery order + 3 consecutive clean eval runs: **16/16, deterministically** --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-07-11 19:31:47 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
pull-request
Mirrored from GitHub Pull Request
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/awesome-llm-apps#14082
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?