| name | aid-ask |
| description | Friendly-named alias of /aid-query-kb -- the optional on-demand Q&A skill. Takes a free-form question and answers it in one pass, grounded in three context sources: the Knowledge Base (.aid/knowledge/), the live codebase, and in-flight AID works (.aid/works/work-*/STATE.md + progress). Returns an answer with source citations. When the available context cannot answer the question, states the gap explicitly and captures it as a Query-Gap entry so it feeds the KB-improvement loop. This file carries no logic of its own -- its full behavior is defined entirely by .cursor/skills/aid-query-kb/SKILL.md, which this skill delegates to.
|
| allowed-tools | Read, Glob, Grep, Agent, Write, Edit |
| argument-hint | <question> — a free-form question about the project |
Project Q&A (alias of /aid-query-kb)
/aid-ask is the friendly-named alias of /aid-query-kb
(.cursor/skills/aid-query-kb/SKILL.md) -- aid-query-kb was renamed FROM
aid-ask in commit cf6cb1af; this file restores the shorter, friendlier
invocation name as a thin alias, per the same alias-is-a-full-directory
convention the generated shortcut catalog uses
(.cursor/aid/templates/shortcut-catalog.yml's Topology note: an alias is a
full, separate .cursor/skills/<name>/SKILL.md directory -- not a redirect
the generator resolves). It is registered in the catalog as
alias_of: aid-query-kb, repurpose: true (never generated or overwritten by
build-shortcut-skills.py, which skips every repurpose: true row -- this
file is hand-authored, like aid-query-kb itself).
This file has no logic of its own. Its entire behavior -- question
classification (trivial vs. broad/expensive), the inline-vs-aid-researcher
dispatch, the answer format, the write-scope constraint (gap-capture only, no
KB/settings/code file ever written), and every other rule -- is defined
exclusively by .cursor/skills/aid-query-kb/SKILL.md. Read that file and
execute it exactly as written, substituting nothing except the invocation
name in any printed usage example (/aid-ask instead of /aid-query-kb):
single-shot, read-only except for the one gap-capture write, grounded in the
Knowledge Base + the live codebase + in-flight .aid/works/work-*/STATE.md works,
cited, no work folder, no state machine of its own. Duplicating that logic
here would let the two skills drift out of sync (feedback precedent: KB
docs never re-derive what another document already owns); this file exists
only so a user who reaches for the shorter, friendlier name still gets the
exact same skill.
Pre-flight
Same as aid-query-kb's own Pre-flight (.cursor/skills/aid-query-kb/SKILL.md § Pre-flight): confirm a question was supplied. If /aid-ask is invoked with
no argument, print:
Usage: /aid-ask <question>
Example: /aid-ask "Which agent tier handles code review?"
Then exit without answering.
Execution
Delegate to .cursor/skills/aid-query-kb/SKILL.md § Execution: answer the question in full -- Step 1 (classify trivial vs. broad/expensive), Step 2a
(trivial: answer inline from KB/codebase reads), Step 2b (broad/expensive:
dispatch aid-researcher read-only), Step 3 (compose the reply, with the
gap-reply shape when context is insufficient), and Step 4 (gap capture: append
a ### Q{N} Query-Gap entry to the resolved backlog's ## Q&A (Pending)
section). Every constraint that file states (§ Constraints) applies here
unchanged: cite sources, never fabricate, write nothing beyond the one
gap-capture append, no .aid/works/work-*/ folder or STATE.md of /aid-ask's own.