Notifications
Clear all
Topic starter
20/08/2026 5:49 pm
TL;DR: Agent harnesses are increasingly managing memory, file reads, tool outputs, and subagents as a governed working set rather than a passive transcript, according to Arize’s analysis of Pi, OpenClaw, Claude Code, and Letta. The security implication is that context pressure now shapes agent reliability, auditability, and the trust boundary around what an agent can safely remember or retrieve.
NHIMG editorial — based on content published by Arize: Context management in agent harnesses: memory, files, and subagents
By the numbers:
- Only 5.7% of organisations have full visibility into their service accounts.
- NHIs outnumber human identities by 25x to 50x in modern enterprises.
- 97% of NHIs carry excessive privileges, increasing unauthorised access and broadening the attack surface.
Questions worth separating out
Q: How should security teams govern memory in AI agents?
A: Treat memory as a privileged state store, not a convenience feature.
Q: Why do long-running AI agents become less reliable over time?
A: Because the context window is finite, so tools, files, and prior turns begin competing with the task itself.
Q: What breaks when an AI agent keeps too much context across troubleshooting runs?
A: It becomes easier for stale hypotheses to shape new actions, which increases false confidence and widens the chance of repeated misdiagnosis.
Practitioner guidance
- Define the authoritative state boundary Document which agent state must remain live in the prompt, which can be summarized, and which must live outside context in files or memory stores.
- Bound tool output before it enters context Set hard limits for file reads, log dumps, and retrieval results, then force pagination or search when outputs exceed the working set.
- Treat subagent creation as delegated access Require explicit scoping for what a child agent can see, which tools it can call, and whether it may inherit parent history.
What's in the full article
Arize's full article covers the operational detail this post intentionally leaves for the source:
- Detailed comparisons of Pi, OpenClaw, Claude Code, and Letta context handling patterns for files, summaries, and subagents
- Specific file-read caps, truncation behaviours, and pagination mechanics that affect implementation decisions
- Compaction triggers, rehydration strategies, and tool-result handling rules that determine what stays visible to the model
- Examples of how different harnesses repair tool-call boundaries and isolate delegated work
👉 Read Arize's analysis of context management in agent harnesses →
Agent harness context management: are your controls keeping up?
Explore further
20/08/2026 6:29 pm
Context management is now an identity governance problem, not just an LLM ergonomics issue. Once an agent harness chooses what to keep, compress, or retrieve, it is determining the practical boundary of the agent’s effective authority. That boundary affects auditability, delegation, and the reliability of downstream decisions. For identity teams, the lesson is that agent memory design belongs in governance review, not only engineering review.
A question worth separating out:
Q: How can teams tell whether context management is actually working?
A: Look for signs that the agent keeps the task coherent after long sessions, retrieves the right file slices on demand, and preserves tool-call boundaries without flooding the prompt. If summaries, pagination, and subagent isolation are traceable and repeatable, the control is doing useful work.
👉 Read our full editorial: Agent harness context management is becoming a governance problem
