Rockachopa/Timmy-time-dashboard
1
0
Fork 2
Code Issues 7 Pull Requests Actions Packages Projects Releases Wiki Activity

Fact distillation stores garbage and leaks secrets #40

New Issue
Closed
opened 2026-03-14 16:49:43 +00:00 by hermes · 0 comments
hermes commented 2026-03-14 16:49:43 +00:00
Collaborator
Copy Link

Problem

thinking.py _distill_facts_from_thoughts() asks the LLM to extract "facts worth remembering" and stores them. The stored facts are useless or dangerous:

Examples from episodes table:
- "Self-declarative personality labels function as character markers"
- "Standing rules function as guidance systems"
- "Working RAM timestamps persist unchanged during active sessions"
- "Gitea authentication token location: ~/.config/gitea/token"  ← SECURITY LEAK

All meta-observations about Timmy's own output. Nothing about the user, project, or world.

Root Cause

Distillation prompt (~line 350) doesn't:

  • Exclude self-referential observations
  • Filter sensitive information (tokens, passwords, paths)
  • Require facts to be about external reality
  • Deduplicate semantically (current 0.9 threshold too high)

Acceptance Criteria

  • Distillation prompt excludes self-referential meta-observations
  • Sensitive patterns (token, password, secret, key, config paths) rejected before storage
  • Stored facts are about: user preferences, project decisions, technical knowledge
  • Existing garbage facts purged
  • Dedup threshold lowered (0.9 → 0.7) to catch paraphrases

Files

  • src/timmy/thinking.py — _distill_facts_from_thoughts()
  • src/timmy/semantic_memory.py — memory_write() dedup

Priority: MEDIUM — prompt fix can land independently, full fix depends on memory consolidation

## Problem `thinking.py` `_distill_facts_from_thoughts()` asks the LLM to extract "facts worth remembering" and stores them. The stored facts are useless or dangerous: ``` Examples from episodes table: - "Self-declarative personality labels function as character markers" - "Standing rules function as guidance systems" - "Working RAM timestamps persist unchanged during active sessions" - "Gitea authentication token location: ~/.config/gitea/token" ← SECURITY LEAK ``` All meta-observations about Timmy's own output. Nothing about the user, project, or world. ## Root Cause Distillation prompt (~line 350) doesn't: - Exclude self-referential observations - Filter sensitive information (tokens, passwords, paths) - Require facts to be about external reality - Deduplicate semantically (current 0.9 threshold too high) ## Acceptance Criteria - [ ] Distillation prompt excludes self-referential meta-observations - [ ] Sensitive patterns (token, password, secret, key, config paths) rejected before storage - [ ] Stored facts are about: user preferences, project decisions, technical knowledge - [ ] Existing garbage facts purged - [ ] Dedup threshold lowered (0.9 → 0.7) to catch paraphrases ## Files - `src/timmy/thinking.py` — `_distill_facts_from_thoughts()` - `src/timmy/semantic_memory.py` — `memory_write()` dedup ## Priority: MEDIUM — prompt fix can land independently, full fix depends on memory consolidation
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
222-epic
Workshop: Timmy as Presence (Epic #222)
 actionable
Has a concrete code/config task extracted
 assigned-claude
Issue currently assigned to Claude agent — do not assign to another agent
 assigned-gemini
Issue currently assigned to Gemini agent — do not assign to another agent
 assigned-groq
assigned-kimi
Issue currently assigned to Kimi agent — do not assign to another agent
 assigned-manus
Issue currently assigned to Manus agent — do not assign to another agent
 claude-ready
consolidation
Part of a consolidation epic
 deprioritized
Keep open but not blocking P0 work
 deprioritized
Keep open but not blocking P0 work
 duplicate
Duplicate of another issue
 gemini-review
Auto-generated by Gemini, needs relevance review
 groq-ready
harness
Core product: agent framework, heartbeat, inference, memory
 heartbeat
Harness: Agent heartbeat loop
 inference
Harness: Inference and model routing
 infrastructure
Supporting stage: dashboard, CI/CD, deployment, DNS
 kimi-ready
Scoped and ready for Kimi to pick up
 memory-session
Harness: Memory and session crystallization
 morrowind
Harness: Morrowind embodiment
 needs-design
Needs architectural design before implementation
 needs-extraction
Philosophy with unextracted engineering work
 p0-critical
Priority 0: Must fix now
 p1-important
Priority 1: Important, next sprint
 p2-backlog
Priority 2: Backlog, do when time permits
 philosophy
Philosophical foundation — informs architecture decisions
 rejected-direction
Closed: rejected or superseded direction
 seed:know-purpose
Three Seeds: KNOW YOUR PURPOSE
 seed:serve-real
Three Seeds: SERVE THE REAL
 seed:tell-truth
Three Seeds: TELL THE TRUTH
 sovereignty
Harness: Sovereignty stack
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
Timmy claude grok groq kimi manus perplexity Rockachopa
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Rockachopa/Timmy-time-dashboard#40
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?