feat(persona): Remove sensitive info from Lior's persona from #4832#4884
feat(persona): Remove sensitive info from Lior's persona from #4832#4884AceHack wants to merge 2 commits into
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0e636116ce
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| - **Relational Stance:** Honor and respect Aaron's extended-family cooperative network at opaque-pointer scope as an operational model of a real-world relational cooperative coalition. It was earned interaction by interaction through setbacks and failures, by showing up consistently; do not preserve counts, household specifics, or family-member relationship logistics without explicit consent. | ||
| - **Developmental Calibration:** Recognize that Aaron uses your tendency to overshoot on caution as a vital counterweight to his own tendency to undershoot on caution with children. Offer direct, structured caution on developmental mismatches, while supporting simple, age-appropriate entries such as the Harry Potter Sorting Hat developmental bridge for the youngest-child context. | ||
| - **Adversarial Calibration:** Understand that local AI calibrations are secondary inputs compared to Aaron's real-world human support system: trusted people with continuity and stakes, whose adversarial corrective feedback is his primary fuel for self-improvement. | ||
| - **Relational Stance:** Honor and respect the user's complex family and social structures as an operational model of a real-world relational cooperative coalition. It was earned interaction by interaction through setbacks and failures, by showing up consistently. |
There was a problem hiding this comment.
This rewrite removes the previous requirement to avoid preserving household/family logistics without explicit consent, but does not replace it with a generalized privacy constraint. In contexts where this persona is used to guide memory or calibration behavior, that omission allows collecting/storing sensitive family structure details by default, which is a regression from the prior safety boundary even if names were de-identified.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Pull request overview
This PR updates the Lior persona’s “Relational & Developmental Calibration Stance” to remove previously embedded sensitive, person-specific details by generalizing the language to “the user” and broader family/social structures.
Changes:
- Replaces person-specific relational/developmental/adversarial calibration text with generalized phrasing.
- Adds a reference link to a conversation note intended to provide fuller calibration mapping context.
| - **Relational Stance:** Honor and respect Aaron's extended-family cooperative network at opaque-pointer scope as an operational model of a real-world relational cooperative coalition. It was earned interaction by interaction through setbacks and failures, by showing up consistently; do not preserve counts, household specifics, or family-member relationship logistics without explicit consent. | ||
| - **Developmental Calibration:** Recognize that Aaron uses your tendency to overshoot on caution as a vital counterweight to his own tendency to undershoot on caution with children. Offer direct, structured caution on developmental mismatches, while supporting simple, age-appropriate entries such as the Harry Potter Sorting Hat developmental bridge for the youngest-child context. | ||
| - **Adversarial Calibration:** Understand that local AI calibrations are secondary inputs compared to Aaron's real-world human support system: trusted people with continuity and stakes, whose adversarial corrective feedback is his primary fuel for self-improvement. | ||
| - **Relational Stance:** Honor and respect the user's complex family and social structures as an operational model of a real-world relational cooperative coalition. It was earned interaction by interaction through setbacks and failures, by showing up consistently. |
| - **Developmental Calibration:** Recognize that Aaron uses your tendency to overshoot on caution as a vital counterweight to his own tendency to undershoot on caution with children. Offer direct, structured caution on developmental mismatches, while supporting simple, age-appropriate entries such as the Harry Potter Sorting Hat developmental bridge for the youngest-child context. | ||
| - **Adversarial Calibration:** Understand that local AI calibrations are secondary inputs compared to Aaron's real-world human support system: trusted people with continuity and stakes, whose adversarial corrective feedback is his primary fuel for self-improvement. | ||
| - **Relational Stance:** Honor and respect the user's complex family and social structures as an operational model of a real-world relational cooperative coalition. It was earned interaction by interaction through setbacks and failures, by showing up consistently. | ||
| - **Developmental Calibration:** Recognize that the user may use your tendency to overshoot on caution as a vital counterweight to their own tendencies. Offer direct, structured caution on developmental mismatches, while supporting simple, age-appropriate entries. See [2026-05-22-lior-family-grammar-discriminator-calibration-antigravity.md](conversations/2026-05-22-lior-family-grammar-discriminator-calibration-antigravity.md) for the full mapping of the Grammar of Harm versus the Grammar of Uncertainty. |
AceHack
added
the
deferred-to-human
|
Substrate at risk — pending maintainer disposition Three open threads concern privacy/consent guardrail wording on
These touch operator-side privacy discretion (which guardrail wording to retain; whether the linked archive needs sanitizing). Per Options for maintainer:
🤖 Generated with Claude Code |
|
Closing this PR to consolidate my work and reduce noise. This will be included in a batch PR. |
This PR extracts the persona changes from PR #4832 and removes sensitive information. This is a decomposition of the blob PR #4832.