You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Hive Commons documentation analysis meta critique #19132
This issue is about what @nate-double-u, head of documentation with the CNCF, likes and would like to see changed about the process kicked off by cncf/techdocs#395 and followed up on through cncf/techdocs#424.
The process looks like it is currently:
\1. maintainer/stake holder opens issue in cncf/techdocs -> 2. cncf/techdocs member triages -> 3. hive drafts the report, implementation plan, and issues -> 4. cncf/techdocs member and stakeholder review -> 5. hive updates
Between steps 3 and 4, I'd like another Hive agent to do the first review (you can see in cncf/techdocs#424 that I did the first review and the Hive responded well to it, but for this first pass I used an LLM to help catch technical errors and analysis holes; this could have been done automatically with an adversarial agent system before the PR was labeled "ready-for-review"
This assessment is still in flight and is going quite well. I'll have more feedback as the process progresses.
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor.
scanner (2026-11-18 20:55 UTC): Scanner triage: this is an in-flight feedback tracker (the author says more feedback is coming), and the requested change, an adversarial review agent between drafting and ready-for-review, is configured per spoke hive rather than in a single code path I can change here. I'm not starting implementation yet, to avoid a speculative change. Once the feedback settles, a follow-up issue naming the specific pipeline stage and the review gate to add would make this actionable.
The PR body says this is partial/non-closing work (Refs/Part of or leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.
Findings: the requested change (an independent agent reviewing a drafted analysis before it is marked ready for review) already exists as the opt-in review.analysis_factcheck pass (#19074, merged in #19080, documented in src/docs/review-swarm.md). It is off by default, so it must be enabled in the TechDocs hive's config. The TechDocs assessment model doc did not mention it; #19150 adds a pointer (docs only). Remaining decision for the operator: enable the pass on that hive and choose independence and ready_policy. Leaving this tracker open for the maintainer's further feedback; any new concrete gap should be filed as its own issue.
Delivered. The requested step, an independent agent reviewing a drafted analysis before it is marked ready for review, is the opt-in review.analysis_factcheck pass (#19080), and the TechDocs assessment workflow doc now points to it with a config example (#19150, merged 21803b9). It is off by default.
To turn it on for a TechDocs hive, add the block documented in src/docs/review-swarm.md under "Independent fact-check for analysis PRs" to that hive's review config: pick the qualifying author_agents or labels, keep independence: model (a different model reviews the draft) and keep ready_policy: human so a person still marks the PR ready. Any new concrete gap in how it behaves should be filed as its own issue. Closing by hand because #19150 used Refs rather than Fixes.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it — check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment /reopen and it will be reopened for you — you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.
Activity
hivecommons-hive commentedon Nov 18, 2026
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor.
PR #19150 covers this issue: #19150
The PR body says this is partial/non-closing work (
Refs/Part ofor leave-open wording), so Hive will not add a closing keyword. GitHub will not show its issue-list PR icon without a closing keyword; the stage comment is the link.hivecommons-hive commentedon Nov 18, 2026
🔓 Claim by @HuntedRaven7 released (lease revoked).
Claim history
HuntedRaven7(contributor on hive hosted-available-oke-11-placeholder-r05x)HuntedRaven7(contributor; lease revoked)clubanderson commentedon Nov 18, 2026
Claimed by the dispatcher session. Reading the feedback, searching for existing work, and verifying against the docs before changing anything.
clubanderson commentedon Nov 18, 2026
Findings: the requested change (an independent agent reviewing a drafted analysis before it is marked ready for review) already exists as the opt-in
review.analysis_factcheckpass (#19074, merged in #19080, documented insrc/docs/review-swarm.md). It is off by default, so it must be enabled in the TechDocs hive's config. The TechDocs assessment model doc did not mention it; #19150 adds a pointer (docs only). Remaining decision for the operator: enable the pass on that hive and chooseindependenceandready_policy. Leaving this tracker open for the maintainer's further feedback; any new concrete gap should be filed as its own issue.📖 docs: point the TechDocs assessment model at the opt-in analysis fa…
clubanderson commentedon Nov 18, 2026
Delivered. The requested step, an independent agent reviewing a drafted analysis before it is marked ready for review, is the opt-in
review.analysis_factcheckpass (#19080), and the TechDocs assessment workflow doc now points to it with a config example (#19150, merged 21803b9). It is off by default.To turn it on for a TechDocs hive, add the block documented in
src/docs/review-swarm.mdunder "Independent fact-check for analysis PRs" to that hive's review config: pick the qualifyingauthor_agentsorlabels, keepindependence: model(a different model reviews the draft) and keepready_policy: humanso a person still marks the PR ready. Any new concrete gap in how it behaves should be filed as its own issue. Closing by hand because #19150 used Refs rather than Fixes.github-actions commentedon Nov 18, 2026
@nate-double-u this issue was closed by @clubanderson as completed.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it — check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment
/reopenand it will be reopened for you — you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.nate-double-u commentedon Nov 18, 2026
Thanks @clubanderson!