fix(scripts): build_standing_set exemplar bias — add declarative-posture patterns#155
Merged
Merged
Conversation
…ure patterns
Pre-fix CONSTRAINT_EXEMPLARS leaned heavily imperative ("never,"
"always," "must," "default to"). Against the real prod vault, the
auto-selected top-10 surfaced 100% engineering rules — career /
lifestyle / posture constraints spanning multi-year load-bearing
facts (runway, recruiter posture, start-date framing, search mode)
did not surface despite being equally durable, because the user
encodes them declaratively ("Brian's stance," "current preference,"
"passive/selective mode") rather than imperatively.
Added 10 declarative-posture exemplars representing the same
constraint class in declarative shape:
- "Brian's stance is correct as-is, posture is by design"
- "current preference is to wait, not to push"
- "his stated preference: keep replies minimal, not desperate"
- "passive / selective mode is correct given runway and pipeline"
- "their silence is information; outreach signals desperation"
- "this decouples cash pressure from outreach push timing"
- "lump sum severance through August is in hand, not biweekly"
- "deliberately niche, not chasing scale or virality"
- "the constraint binding decisions is preference, not necessity"
- "soft target for start date preserves negotiating leverage"
Exemplar list 22 → 30. Imperative / declarative split now roughly
balanced (5 imperative-shape / 7 declarative-shape on inspection).
ROADMAP audit-finding follow-up per
[[feedback_audit_findings_become_roadmap_followups]] — surfaced during
salience-tier Phase 0 validation attempts (handoff 2526). Operator
should re-run scripts/salience_phase0.sh against a fresh snapshot to
verify the bias fix surfaces career-context memories alongside the
engineering rules in the top-10.
No tests depend on the exemplar list (verified via grep); suite stays
at 836 passing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
4 tasks
cipher813
added a commit
that referenced
this pull request
May 22, 2026
Seals the 2026-05-22 substrate arc: - #153 capture-attention Phase A (recurrence-weighted preserve+relate+boost) - #154 salience-tier Phase 1 (first-class standing tier, +3 MCP tools) - #155 build_standing_set.py exemplar bias fix Both new feature paths gated default-off (CAPTURE_ATTENTION_ENABLED, STANDING_TIER_ENABLED) — operator flips per the soak workflow. Post-merge ritual: - tag v0.7.0rc1 + GitHub Release - twine upload (dist/mnemon-memory-0.7.0rc1.{tar.gz,whl}) - mnemon upgrade web --app-name mnemon-memory --mnemon-version 0.7.0rc1 - mnemon doctor 7/7 against live remote - operator promotes 5 career memories + flips MNEMON_STANDING_TIER_ENABLED for ≥1 week soak Suite 836 passing. mnemon --version returns 0.7.0rc1. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The pre-fix
CONSTRAINT_EXEMPLARSlist inscripts/build_standing_set.pyleaned heavily imperative ("never," "always," "must," "default to"). The 2026-05-22 finding during salience-tier Phase 0 validation attempts: against the real prod vault the auto-selected top-10 was 100% engineering rules. Career / lifestyle / posture constraints spanning multi-year load-bearing facts (runway, recruiter posture, start-date framing, job-search mode) did not surface despite being equally durable — because the user encodes them declaratively ("Brian's stance," "current preference," "passive/selective mode") rather than imperatively.Adds 10 declarative-posture exemplars representing the same constraint class in declarative shape. Exemplar list 22 → 30. Imperative / declarative split now roughly balanced.
Driver
ROADMAP audit-finding follow-up per
feedback_audit_findings_become_roadmap_followups. Filed when the Phase 0 A/B attempt revealed the bias.Changes
scripts/build_standing_set.py: appended 10 newCONSTRAINT_EXEMPLARS:Composability
memory_promoteMCP tool. This PR helps the Phase 0 env-var fallback path (used as operator override when not flipping the Phase 1 flag) and the auto-suggestion loop (scripts/salience_phase0.sh score) surface the right candidates by default.Test plan
python -c \"import scripts.build_standing_set\")scripts/salience_phase0.sh snapshot && scripts/salience_phase0.sh scoreagainst a fresh vault snapshot; verifies that the top-10 surfaces at least 3 career-context memories alongside the engineering rules🤖 Generated with Claude Code