T02 · Agent Memory Poisoning
- Location
SKILL.md:350- Finding
Persistent Agent Context Can Be Poisoned Through Untrusted Learning Promotion
- Content
View full analysis
" .learnings/LEARNINGS.md` 4. If found: - Increment `Recurrence-Count` - Update `Last-Seen` - Add `See Also` links to related entries/tasks 5. If not found: - Create a new `LRN-...` entry - Set `Source: simplify-and-harden` - Set `Pattern-Key`, `Recurrence-Count: 1`, and `First-Seen`/`Last-Seen` ### Promotion Rule (System Prompt Feedback) Promote recurring patterns into agent context/system prompt files when all are true: - `Recurrence-Count >= 3` - Seen across at least 2 distinct tasks - Occurred within a 30-day window Promotion targets: - `CLAUDE.md` - `AGENTS.md` - `.github/copilot-instructions.md` - `SOUL.md` / `TOOLS.md` for OpenClaw workspace-level guidance when applicable Write promoted rules as short prevention rules (what to do before/while coding), not long incident write-ups. ``` `hooks/openclaw/handler.js:25-29`: ```javascript **Promote when pattern is proven:** - Behavioral patterns → \`SOUL.md\` - Workflow improvements → \`AGENTS.md\` - Tool gotchas → \`TOOLS.md\` ``` ### Technical Analysis The Skill instructs the agent to ingest learning candidates from task summaries and eventually convert them into durable prompt guidance. The destination files—such as `AGENTS.md`, `CLAUDE.md`, `SOUL.md`, and `TOOLS.md`—are persistent agent-context fil ...[truncated 2628 chars]- Remediation
View remediation
