searxng-mcp-go/docs/agents/triage-labels.md
XLion d349629225
Sync triage-labels.md with actual GitHub labels
Add 11 new entries to Known Labels that were created on the
repo but missing from docs/agents/triage-labels.md: technical-debt,
code-quality, performance, testing, maintainability,
needs-investigation, mcp, searxng, error-handling,
context-timeout-retry, and joke.

Also align description wording with actual label descriptions
from GitHub.
2026-06-17 02:34:22 +08:00

3.8 KiB

Triage Labels

Issue labels live in GitHub. Always read the current list with gh label list before doing label-sensitive work.

Known Labels

This table documents the expected label set, but GitHub is authoritative. If gh label list differs from this table, use the live GitHub labels and mention the documentation drift.

Label Meaning Agent use
bug Something is not working Agents may apply when the issue describes a defect.
documentation Documentation improvement Agents may apply for docs-only work.
duplicate Existing issue already covers this Agents may suggest or apply only when the duplicate issue is clearly identified.
enhancement New feature or request Agents may apply for non-bug improvements.
good first issue Good for newcomers Agents may suggest; humans decide final suitability.
help wanted Extra attention is needed Agents may suggest; humans decide final suitability.
invalid This doesn't seem right Agents may suggest; humans should make final closure decisions.
question Further information is requested Agents may apply when asking the reporter for more information.
wontfix This will not be worked on Human decision label; agents may recommend but should not apply without explicit human approval.
security Security problems Agents may apply for security-relevant issues, but do not disclose unvalidated exploit details beyond what is needed for triage.
technical-debt Technical debt and design cleanup Agents may apply for refactoring and design improvement issues.
code-quality Code quality and maintainability Agents may apply for code quality concerns.
performance Performance or resource-efficiency concerns Agents may apply for performance-related issues.
testing Tests, coverage, and regression protection Agents may apply for testing improvements.
maintainability Maintainability, structure, and refactoring candidates Agents may apply for maintainability concerns.
needs-investigation Needs deeper investigation before implementation Agents may apply when the issue needs more research.
mcp Model Context Protocol integration Agents may apply for MCP-related changes.
searxng SearXNG integration and search behavior Agents may apply for SearXNG-specific changes.
error-handling Error taxonomy, wrapping, and reporting Agents may apply for error-related changes.
context-timeout-retry Context propagation, timeouts, cancellation, and retry behavior Agents may apply for context/timeout/retry changes.
joke It is joke Agents may apply for humorous or off-topic entries.

Human-Only Decision Labels

The following labels are operator decision states. Agents must not apply, remove, or change these labels unless the user explicitly asks for that exact label operation. Agents may mention that one of these labels seems appropriate in a comment or report.

Label Meaning
accepted The claims are accepted and approved for action.
needs-explain The human operator does not understand the claim or is not convinced yet.
rejected The claims are explicitly rejected.

Role Mapping

Some skills use generic role names. Map them to this repo's labels as follows:

Generic role Use in this repo
needs-triage Do not apply a label automatically; leave unlabeled or use a topical label such as bug, enhancement, documentation, or security.
needs-info Use question when asking for more information.
ready-for-agent No direct label. Mention readiness in the issue/comment instead.
ready-for-human No direct label. Mention that human judgment is needed instead.
wontfix Use wontfix only with explicit human approval.