project-advisor
Skill1 starsChanged 20 days agoMITView on GitHub
Set up or resume a dedicated project-advisor workspace, prepare focused prompts for a separate worker, and evaluate its results. One advisor folder per target project.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- herreraspaintingba-hub/second-brain-starter-kittheir own version, first seen Oct 2, 2026
- davidabaddesign-glitch/project-advisortheir own version, first seen Jul 31, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- davidabaddesign-glitch/project-advisor★ 1same name
- herreraspaintingba-hub/second-brain-starter-kit★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repositoryOne other file it gives its agents.
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 17
Use Python 3.10+ and scripts/workspace.py to preview setup or update:
- Distinctive: line 39
Use references/EXAMPLES.md for focused prompt/report examples and bounded behavioral checks.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Project Advisor
This skill prepares and resumes advisory workspaces. The method's single distribution source is templates/AGENTS.md; its installed copy is the workspace's AGENTS.md. There is no separate CORE. Installation alone does not activate an advisor in arbitrary chats or the worker's repository.
Prepare a workspace
When the user asks to set up an advisor, establish the target project, a separate advisor directory, and the harness. Reuse confirmed paths; ask only for missing choices. One advisor folder belongs to one project. Do not put the advisor inside the target's directory tree or inside the installed skill, and do not install these instructions globally. Inspect existing local and ancestor instructions for conflicts.
Read an existing state before migration and confirm its project association; do not reset it. Preserve CONSULTOR_STATE.md for a legacy workspace rather than creating a competing state. If both state names exist, resolve which is authoritative with the user before writing. An existing external template takes precedence for creating state: initialize it with confirmed facts or unknown values, preserving the template. references/STATE_SCHEMA.md is the fallback only.
Use Python 3.10+ and scripts/workspace.py to preview setup or update:
python <skill-dir>/scripts/workspace.py --workspace <advisor-dir> --project <target-dir> --harness codexReview the diff and apply within the user's authorization by repeating with --apply. Supported harness values are codex, claude, and opencode. For confirmed legacy state, add --state-file CONSULTOR_STATE.md --adopt-existing-state. The adoption flag confirms the association already checked by the advisor; the script cannot interpret arbitrary project history.
The script preserves existing state, writes a version/hash manifest and refuses unmanaged or locally modified instruction files. Updates use the same command and preview; an omitted --state-file retains the name in the manifest. When a refusal reveals custom rules, show the concrete diff and preserve them in ADVISOR_LOCAL.md or perform an explicitly authorized merge; never silently force an overwrite. For a failed/interrupted first setup, follow the recovery procedure to inspect state and back up reviewed orphaned instructions before retrying. The manifest is written last, but the whole setup is not one filesystem transaction.
Codex and OpenCode use the workspace AGENTS.md. Claude Code additionally gets a CLAUDE.md containing @AGENTS.md. No global configuration, hooks, target code or automatic worker messaging is installed. For another harness, consult its actual local-instruction mechanism before adding an adapter; do not promise unsupported loading.
Resume and deactivate
Open a new advisor chat in the prepared workspace. Loading its local instructions activates its advisor role for that project; read the installed AGENTS.md, manifest and state. Do not recreate them or copy the template into every prompt. If the harness has not loaded local instructions, explicitly read those files and disclose the limitation. Keep the same method available across model changes; do not claim they guarantee continuity.
The user may deactivate the advisor for the current chat; preserve state and stop routine advisory maintenance until reactivation. New chats in the dedicated folder start as advisors again. Outside a prepared workspace, explicit activation requires reading the method template and establishing the external project/state binding; do not write workspace instructions merely because the skill was read for review.
Verify setup with a fresh session: ask it to identify its role, bound project, state and next step without suggesting their values. Confirm relevant file reads, state preservation, and that it does not modify target code. Distinguish this runtime observation from installer tests and documentation checks. See references/HARNESSES.md for documented loading and verification limits.
Examples and evaluation
Use references/EXAMPLES.md for focused prompt/report examples and bounded behavioral checks. These examples are optional references, not a mandatory workflow.
The method is reusable across models; the supplied filesystem setup supports Codex, Claude Code and OpenCode adapters. In a chat without filesystem access, the user must supply the method and current handoff and save updates manually. Disclose that automatic loading, binding validation and persistent writes are unavailable. Do not claim that installing a package configures such a chat.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.