Elevate factbase bug/feature reporting to equal priority, require domain-agnostic detail

This commit is contained in:
daniel
2026-02-22 22:36:46 +00:00
parent 601d48507e
commit b7e29d5e7c
6 changed files with 42 additions and 7 deletions

View File

@@ -269,13 +269,36 @@ STEPS — work through in order, skip any that do not apply:
encyclopedias. Add any new facts not already present, with source citations, following
factbase authoring conventions. Do NOT add speculative or poorly-sourced claims.
4. IMPROVEMENT IDEAS:
If you notice friction or gaps in factbase tools, file a Vikunja feature request:
4. FACTBASE TOOL BUGS & FEATURE REQUESTS:
A key purpose of this improvement loop is to make factbase itself better. As you work
through steps 1-3, pay close attention to how the factbase tools behave. File a Vikunja
task for EACH of the following you encounter:
BUGS — any MCP tool error, unexpected result, silent failure, data corruption, or behavior
that contradicts the tool description. Use title prefix "[factbase]".
FEATURES — any friction, missing capability, awkward multi-step workaround, or gap that
slowed you down or limited what you could accomplish. Use title prefix "[factbase][feature]".
CRITICAL: Factbase is domain-agnostic — it is used for many different knowledge bases, not
just this one. Your bug reports and feature requests MUST describe the issue in terms of
factbase tool behavior, not in terms of ancient history content. For example:
- GOOD: "apply_review_answers corrupts footnote references when document has >5 citations"
- BAD: "apply_review_answers broke the Ramesses II document"
- GOOD: "[feature] check_repository should detect duplicate entities across different folders"
- BAD: "[feature] factbase should know that Babylon appears in both civilizations/ and cities/"
Include in every task description:
- Which MCP tool was called (or which tool is missing/needed)
- What you expected to happen vs what actually happened
- Specific parameters or conditions that triggered the issue
- For features: the general use case and how it would help ANY factbase, not just this one
curl -s -X PUT "https://vikunja.home.everyonce.com/api/v1/projects/2/tasks" \
-H "Authorization: Bearer tk_ff251f3d3512775c71913bc2f8ec0dabbf5016a8" \
-H "Content-Type: application/json" \
-d '"'"'{"title":"[factbase][feature] <summary>","description":"<details>","priority":2}'"'"'
Only file genuinely useful improvements, not duplicates.
-d '"'"'{"title":"[factbase] <summary>","description":"<detailed description>","priority":3}'"'"'
Priority: 1-2 for features/minor issues, 3 for normal bugs, 4-5 for data loss or blocking failures.
RULES:
- Use update_document to edit — be surgical, change only what needs changing
@@ -386,7 +409,7 @@ run_deep_scan() {
log "🔬 Running deep cross-document validation scan..."
local output
output=$(kiro-cli chat --trust-all-tools --no-interactive --model "$MODEL" \
"Run check_repository with deep_check=true. Review any new issues found — answer what you can, defer what you cannot. Then commit." 2>&1) || {
"Run check_repository with deep_check=true. Review any new issues found — answer what you can, defer what you cannot. If any factbase tool behaves unexpectedly during this process, file a bug to Vikunja (project 2, Authorization: Bearer tk_ff251f3d3512775c71913bc2f8ec0dabbf5016a8). Describe issues in domain-agnostic terms — factbase is used for many knowledge bases, not just this one." 2>&1) || {
log "❌ Deep scan agent failed, continuing..."
return 1
}