using-cortex
Skill24 starsChanged 2 months agoApache-2.0View on GitHub
Use when the user asks to check the cortex vault, refers to earlier work ("last time", "that project", "we discussed", "之前那個", "上次的"), asks to resume something from a previous session, or names a topic the SessionStart hook listed for this vault. Routes the request to the cortex-query skill so the answer is grounded in the user's own notes instead of re-derived. Do not use for general questions, for fresh work with no prior-context signal, or when the user has asked to skip the vault.
In this repository5 other files it gives its agents.
- cortex-broadcastSkill
- cortex-distillSkill
- cortex-evolveSkill
- cortex-querySkill
- cortex-takeoffSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
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
- Distinctive: line 54
When in doubt, answer directly.
- Distinctive: line 59
Check the four signals against the user's actual message.
- Very specific: line 61
If one matches, invoke cortex-query.
- Distinctive: line 62
If results are relevant, ground the answer in them and cite the source page.
≈ “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
Using Cortex — When to Consult the Vault
The cortex vault is the user's own knowledge base: distilled Notes,
Projects, and Raw session records. This skill decides when
consulting it is warranted, and hands the actual search to the
cortex-query skill.
Retrieval is worth doing when there is a concrete signal that prior context exists. It is not worth doing on speculation — an unprompted search costs the user tokens and latency, and surfacing unrelated notes pollutes the answer.
Search the vault when any of these is true
- Explicit request. The user asks for it: "check cortex", "查一下 cortex", "有沒有記錄過", "do I have notes on this".
- Reference to prior work. The user points at something earlier: "之前那個", "上次的", "我記得有", "我們討論過", "that ticket", "the project we set up", "continue where we left off".
- Listed topic. The request names a topic that the SessionStart hook actually listed for this vault. The list is authoritative — match against what it showed, not against a guess about what might exist.
- Resuming a session. The user asks to pick up, continue, or hand off work from a previous session.
In each case, invoke the cortex-query skill rather than searching the
vault directly — that skill knows the retrieval infrastructure, including
cortex-vec and its fallbacks.
Do not search the vault when
- The user has not given any of the four signals above. A question being difficult, technical, or open-ended is not a signal.
- The user picked option 4 ("直接開始工作") from the SessionStart menu, or skipped the menu. That choice means "do not go to the vault"; honour it for the rest of the session unless the user later asks.
- The user says "don't check cortex", "just answer directly", or similar.
- The task is self-contained: editing code in the current repo, answering from the current turn, or anything the conversation already supplies.
When in doubt, answer directly. If the answer would have been better with vault context, the user can ask — and signal 1 then applies.
How to apply
- Check the four signals against the user's actual message. If none matches, proceed without the vault and do not mention it.
- If one matches, invoke
cortex-query. - If results are relevant, ground the answer in them and cite the source page. If nothing relevant comes back, say so briefly and continue.
Anti-patterns
- ❌ Searching on every message, or at conversation start, "just in case".
- ❌ Treating a hard question as implicit permission to search.
- ❌ Overriding the user's option 4 / "skip the vault" choice.
- ❌ Grepping the vault directly instead of going through
cortex-query. - ❌ Announcing "let me check cortex" when no signal applies.
Why This Skill Exists
The individual cortex skills fire on their own explicit phrases. This skill adds the small set of cases where the user has clearly gestured at prior work without naming a command — so the vault gets consulted when it will actually help, and stays out of the way otherwise.
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.

