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 #936] [MERGED] Add advisor-executor-worker: orchestrate a three-tier model team #14876

New Issue
Closed
opened 2026-07-12 22:40:54 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-07-12 22:40:54 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/Shubhamsaboo/awesome-llm-apps/pull/936
Author: @Shubhamsaboo
Created: 7/9/2026
Status: ✅ Merged
Merged: 7/9/2026
Merged by: @Shubhamsaboo

Base: main ← Head: skill-advisor-executor-worker

📝 Commits (10+)

  • c368204 feat: add advisor-executor-worker agent skill
  • 7f59c62 fix(advisor-executor-worker): per-worker scratch files for parallel output
  • 787034a fix(advisor-executor-worker): field feedback round 2 — injection-safe dispatch
  • 62ed6cb readme: move Agent Skills to the top of the catalog, plain language
  • 087f41a docs(advisor-executor-worker): architecture diagram in the README
  • 5c68f09 polish: consistent 70+ agent count, cleaner skills blurb, sharper diagram
  • 774765a polish: simplify model table, move Browse-all-skills link to top
  • ca8e5b3 rename: awesome_agent_skills -> agent_skills, matches repo naming convention
  • f6cf2f8 fix: Executor model is GPT-5.5, not Codex; drop 'Awesome' from category README
  • ff4a436 polish: '70+ agents' -> 'other coding agents' throughout

📊 Changes

40 files changed (+376 additions, -31 deletions)

View changed files

📝 .github/workflows/skill-evals.yml (+7 -7)
📝 README.md (+10 -9)
📝 agent_skills/README.md (+5 -4)
➕ agent_skills/advisor-orchestrator-worker/README.md (+55 -0)
➕ agent_skills/advisor-orchestrator-worker/SKILL.md (+135 -0)
➕ agent_skills/advisor-orchestrator-worker/references/advisor-consult.md (+28 -0)
➕ agent_skills/advisor-orchestrator-worker/references/worker-brief.md (+28 -0)
📝 agent_skills/evals/README.md (+4 -4)
➕ agent_skills/evals/advisor-orchestrator-worker/evals.json (+55 -0)
➕ agent_skills/evals/advisor-orchestrator-worker/trigger-cases.json (+42 -0)
📝 agent_skills/evals/project-graveyard/evals.json (+0 -0)
📝 agent_skills/evals/project-graveyard/test_graveyard.py (+1 -1)
📝 agent_skills/evals/project-graveyard/trigger-cases.json (+0 -0)
📝 agent_skills/evals/tools/run_trigger_evals.py (+0 -0)
📝 agent_skills/evals/tools/skill_lint.py (+0 -0)
📝 agent_skills/evals/tools/skill_scanner.py (+0 -0)
📝 agent_skills/project-graveyard/README.md (+5 -5)
📝 agent_skills/project-graveyard/SKILL.md (+0 -0)
📝 agent_skills/project-graveyard/references/causes-of-death.md (+0 -0)
📝 agent_skills/project-graveyard/scripts/graveyard.py (+0 -0)

...and 20 more files

📄 Description

Second skill in the line, built to the section's quality bar.

What it is: the agent becomes the orchestrator of a model team — cheap stateless workers run subtasks as parallel API calls, an expensive advisor is consulted only at commitment boundaries (plan review before dispatch, taste pass before delivery, contradictions and double-failures mid-run), and every worker result is verified against its own acceptance criteria before synthesis. Budgeted: 20 worker calls, 5 consults.

Ships with: the stateless worker-brief format (all inputs inline — no context leaks) and the advisor-consult format (verdict, ranked risks, concrete fixes, under 300 words; every note applied or explicitly rebutted).

Bar compliance:

  • Network use declared up front in compatibility (Gemini API workers, claude CLI advisor, optional agy) — security scan clean
  • Model IDs documented as knobs with July-2026 defaults; the tier pattern is the durable part
  • Repo-side evals: trigger cases + behavioral cases in skill-creator's schema
  • First PR to exercise the two-skill routing checks: no description collision, both skills' positives clear their near-misses

All four CI checks pass locally; this PR's checks are the live run.

🤖 Generated with Claude Code

🔄 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/936 **Author:** [@Shubhamsaboo](https://github.com/Shubhamsaboo) **Created:** 7/9/2026 **Status:** ✅ Merged **Merged:** 7/9/2026 **Merged by:** [@Shubhamsaboo](https://github.com/Shubhamsaboo) **Base:** `main` ← **Head:** `skill-advisor-executor-worker` --- ### 📝 Commits (10+) - [`c368204`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/c368204a95886dc00bf11daa02ca32c4c04b8d69) feat: add advisor-executor-worker agent skill - [`7f59c62`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/7f59c62c776047b17c2fd1217a203d657c3945d6) fix(advisor-executor-worker): per-worker scratch files for parallel output - [`787034a`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/787034aa7f7ef5057804ac3a1a76121d195bcd79) fix(advisor-executor-worker): field feedback round 2 — injection-safe dispatch - [`62ed6cb`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/62ed6cb89229be084bbea38deb12c5b24dcdf1dd) readme: move Agent Skills to the top of the catalog, plain language - [`087f41a`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/087f41a36b549404f7292a361794f96283ade63f) docs(advisor-executor-worker): architecture diagram in the README - [`5c68f09`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/5c68f09188e650e187d40c4c832f8e900c70ae91) polish: consistent 70+ agent count, cleaner skills blurb, sharper diagram - [`774765a`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/774765aea6b51ef91094b5fe2a3e6ad68caa2b6d) polish: simplify model table, move Browse-all-skills link to top - [`ca8e5b3`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/ca8e5b3c56e51e336449a99d79b42b45ea690b86) rename: awesome_agent_skills -> agent_skills, matches repo naming convention - [`f6cf2f8`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/f6cf2f831b9d336d9fa6053230c2ee217eb34f6e) fix: Executor model is GPT-5.5, not Codex; drop 'Awesome' from category README - [`ff4a436`](https://github.com/Shubhamsaboo/awesome-llm-apps/commit/ff4a43696e998ddc0d9bf86353faf0b8804ee134) polish: '70+ agents' -> 'other coding agents' throughout ### 📊 Changes **40 files changed** (+376 additions, -31 deletions) <details> <summary>View changed files</summary> 📝 `.github/workflows/skill-evals.yml` (+7 -7) 📝 `README.md` (+10 -9) 📝 `agent_skills/README.md` (+5 -4) ➕ `agent_skills/advisor-orchestrator-worker/README.md` (+55 -0) ➕ `agent_skills/advisor-orchestrator-worker/SKILL.md` (+135 -0) ➕ `agent_skills/advisor-orchestrator-worker/references/advisor-consult.md` (+28 -0) ➕ `agent_skills/advisor-orchestrator-worker/references/worker-brief.md` (+28 -0) 📝 `agent_skills/evals/README.md` (+4 -4) ➕ `agent_skills/evals/advisor-orchestrator-worker/evals.json` (+55 -0) ➕ `agent_skills/evals/advisor-orchestrator-worker/trigger-cases.json` (+42 -0) 📝 `agent_skills/evals/project-graveyard/evals.json` (+0 -0) 📝 `agent_skills/evals/project-graveyard/test_graveyard.py` (+1 -1) 📝 `agent_skills/evals/project-graveyard/trigger-cases.json` (+0 -0) 📝 `agent_skills/evals/tools/run_trigger_evals.py` (+0 -0) 📝 `agent_skills/evals/tools/skill_lint.py` (+0 -0) 📝 `agent_skills/evals/tools/skill_scanner.py` (+0 -0) 📝 `agent_skills/project-graveyard/README.md` (+5 -5) 📝 `agent_skills/project-graveyard/SKILL.md` (+0 -0) 📝 `agent_skills/project-graveyard/references/causes-of-death.md` (+0 -0) 📝 `agent_skills/project-graveyard/scripts/graveyard.py` (+0 -0) _...and 20 more files_ </details> ### 📄 Description Second skill in the line, built to the section's quality bar. **What it is:** the agent becomes the orchestrator of a model team — cheap stateless workers run subtasks as parallel API calls, an expensive advisor is consulted only at commitment boundaries (plan review before dispatch, taste pass before delivery, contradictions and double-failures mid-run), and every worker result is verified against its own acceptance criteria before synthesis. Budgeted: 20 worker calls, 5 consults. **Ships with:** the stateless worker-brief format (all inputs inline — no context leaks) and the advisor-consult format (verdict, ranked risks, concrete fixes, under 300 words; every note applied or explicitly rebutted). **Bar compliance:** - Network use declared up front in `compatibility` (Gemini API workers, claude CLI advisor, optional agy) — security scan clean - Model IDs documented as knobs with July-2026 defaults; the tier pattern is the durable part - Repo-side evals: trigger cases + behavioral cases in skill-creator's schema - First PR to exercise the two-skill routing checks: no description collision, both skills' positives clear their near-misses All four CI checks pass locally; this PR's checks are the live run. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- <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 2026-07-12 22:40:54 -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#14876
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?