hermes - 💡(How to fix) Fix [Feature]: Allow null-result outcomes in bg-review's skill update prompt

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Fix Action

Fix / Workaround

"Review the conversation above and update the skill library only when a clear signal warrants it (user corrections, frustration markers, novel patterns, outdated content, missing pitfalls). Prefer patching an existing skill over creating a new file. If no signal fired this session, respond exactly 'No update warranted.' — null results are valid."

RAW_BUFFERClick to expand / collapse

Problem or Use Case

The bg-review prompt at agent/background_review.py:48-51 (the _SKILL_REVIEW_PROMPT constant) instructs the review agent:

"Review the conversation above and update the skill library. Be ACTIVE — most sessions produce at least one skill update, even if small. A pass that does nothing is a missed learning opportunity, not a neutral outcome."

This frames inaction as failure. Predictable consequence: the review model produces edits even when no clear signal warrants one. Concrete case: a daily-briefing cron run with no user correction, no novel pattern, no skill consulted — bg-review still authored two complementary-but-overlapping reference files in 9 seconds. The vague target ("at least one skill update, even if small") combined with the "missed opportunity" framing manufactures noise.

For comparison, the parallel _MEMORY_REVIEW_PROMPT already allows honest null results: "If nothing is worth saving, just say 'Nothing to save.' and stop." The skill review prompt does not have an equivalent escape hatch.

Proposed Solution

Reframe the skill review prompt to allow honest null outcomes, e.g.:

"Review the conversation above and update the skill library only when a clear signal warrants it (user corrections, frustration markers, novel patterns, outdated content, missing pitfalls). Prefer patching an existing skill over creating a new file. If no signal fired this session, respond exactly 'No update warranted.' — null results are valid."

Keep the existing "preference order" (update loaded skill → update umbrella → add support file) as guidance for when to act, just not as pressure to always act.

Alternatives Considered

  • Leave the prompt and raise creation_nudge_interval default — fires less often, but each firing still has the action bias.
  • Make the prompt configurable via skills.bg_review_prompt_override — gives users an escape hatch but does not fix the bundled default.

Feature Type: Configuration option Scope: Small

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING