Review memory/checkpoint provenance

1. 01

   Concepts & system design

   60 min· read · map the mechanismtimer

   R24 — LangGraph checkpointing

   R41 — Agentic memory/state risks

   Read the assigned sections; map the mechanism, threat model and trade-offs.

2. 02

   Project implementation

   120 min· build · integrate · debugtimer

   Audit P3 LangGraph checkpoint provenance, tenant binding, replay prevention, retention TTLs, deletion, and D11 cryptographic hash-chain integrity across every agent step and tool side effect.

   Build, integrate, debug and improve the increment in the active project.

3. 03

   Security lab

   60 min· attack your own systemtimer

   Test cross-tenant checkpoint restoration, stale state replay, poisoned memory injection (D10), and database audit-log tampering (D11).

   Test the abuse case, failure path, authorization bypass or resource limit.

4. 04

   Verification & evidence

   30 min· test · record evidencetimer

   P3 checkpoint isolation and D11 tamper-evidence verification report.

   Add or run regression tests; record the 9-point evidence record.

5. 05

   Deliberate practice & Socratic review

   30 min· explain aloud · update gapstimer

   Answer the 8 Socratic questions without notes: What? Why? How? What if? Why not? How can I prove it? Where can it break? What have I learned?

   Re-explain the hardest concept aloud, inspect one edge case, update the gap log.

6. total5 h

• R24LangGraph Documentation↗ docs.langchain.com

  State graphs, durable checkpointing, human-in-the-loop interrupts, retries, and resumability.

• R41OWASP Agentic AI Resources↗ genai.owasp.org

  Agentic threat taxonomy: tool misuse, excessive agency, memory poisoning, inter-agent trust, and human oversight.

Read only the named sections needed for today's task. Reading sits inside the 60-minute concepts block; it is not extra time.

• docsDocs by LangChain
  Persistence

  Defines the checkpointers-vs-stores split and thread-scoped state, which is the object you must bind to tenant and provenance before auditing replay and retention TTLs.

  R24checked 2026-11-11
• videoTyrannoFlow
  LangGraph Workflow Checkpointing and Interrupts - Part 1/3

  Implements checkpointing with a saver keyed to a thread_id, showing concretely what a persisted step snapshot contains and therefore what an attacker could replay.

  R24checked 2026-11-11
• videoProduct Security Podcast · 6:27
  Memory Poisoning in AI Security

  Explains how a poisoned write persists across sessions and contaminates later retrievals, the exact failure mode your D10 scrubber and checkpoint provenance checks must detect.

  R41checked 2026-11-11
• articlePalo Alto Unit 42
  When AI Remembers Too Much - Persistent Behaviors in Agents' Memory

  Documents indirect prompt injection surviving into long-term memory and enabling history exfiltration, which motivates hashing each memory write into the D11 trajectory ledger.

  R41checked 2026-11-11
• docsDocs by LangChain
  Checkpointers

  Details that a checkpoint is written at every superstep with a monotonic ID, giving you the per-step anchor points for hash-chaining and deletion verification.

  R24checked 2026-11-11

A lab is complete only when it produces evidence. Installing PyRIT, Garak, Presidio or TransformerLens is not completion. Saved on this device only.