zippolyon

9 ms·

1.
by zippolyon 2mo ago
You mentioned that the system supports interventions and counterfactuals. I am curious: does it simply recalculate a predicted probability based on the intervention, or does it separately log the target, current context, intervention, actua
2.
by zippolyon 6mo ago
This is the exact problem that keeps us up at night. We ran a controlled experiment: same AI agents, same task, two conditions. Without runtime enforcement, our CMO agent fabricated an audit record — invented a governance event that n
3.
by zippolyon 6mo ago
Interesting premise — agree that agents need runtime infrastructure, not just frameworks. One dimension we've been exploring: runtime governance. Even with a good runtime, agents can fabricate compliance records, silently drop ta
4.
What Happens When You Tell AI Agents the Rules but Don't Enforce Them(github.com)
2 points by zippolyon 6mo ago|0 comments
5.
by zippolyon 6mo ago
The bundle K9 seals at execution time covers most of what you described: X_t captures agent identity, session ID, hostname and PID at the moment of execution; Y _t hashes the constraint version in force; each record chains via SHA256 prev_h
6.
by zippolyon 6mo ago
Fair point — it's Python-based if you use the @k9 decorator directly. For Claude Code specifically, the hook works regardless of what language the agent is written in, because it intercepts at the tool call level via .claude/setti
7.
by zippolyon 6mo ago
Not Claude Code specific — @k9 decorator works with any Python agent (LangChain, AutoGen, CrewAI). For Claude Code it hooks via .claude/settings.json, zero code changes. For pi-agent: if it's Python-based, yes it works. https:&#x
8.
I used my AI auditing tool to audit itself – and it found two real problems(github.com)
1 points by zippolyon 6mo ago|1 comments
9.
by zippolyon 6mo ago
K9 Audit records every AI agent action as a CIEU five-tuple (context / action / intent / outcome / assessment), SHA256 hash-chained locally. Zero tokens consumed, no external servers. I wondered: does K9 actually do what
10.
by zippolyon 6mo ago
The hesitation about log upload is exactly why K9 Audit works differently — local by default, SHA256 hash-chained, zero data leaves your machine unless you explicitly configure a sync endpoint. pip install k9audit-hook and drop one JSON fil
11.
by zippolyon 6mo ago
The billing surprise problem is real, but there's a deeper one: exit code 0 with silent deviations. K9 Audit complements OTel — instead of just tracing tokens and latency, it records intent vs actual outcome for every tool call as a CI
12.
by zippolyon 6mo ago
Running 8 parallel agents is exactly when silent deviations get expensive — one agent writes to the wrong path and exit code 0 hides it. K9 Audit drops into .claude/settings.json and records every tool call as a cryptographic evidence
13.
by zippolyon 6mo ago
Great work on the session analytics. The "error cascade in first 2 minutes predicts abandonment" finding is exactly the kind of signal that causal auditing can act on. We built K9 Audit for the complementary problem: not just when
14.
by zippolyon 6mo ago
Nice work on the gitagent port. K9 Audit drops straight into .claude/settings.json — zero-config audit layer, works alongside gstack. https://github.com/liuhaotian2024-prog/K9Audit
15.
by zippolyon 6mo ago
Nice work on the gitagent port. If anyone runs this agent autonomously, K9 Audit drops straight into .claude/settings.json as a zero-config audit layer — records every tool call as a cryptographic evidence chain, flags silent deviation
16.
by zippolyon 6mo ago
Not Claude Code specific — works with any Python agent via a one-line decorator. LangChain, AutoGen, CrewAI, or anything custom. Claude Code just gets a zero-config hook via .claude/settings.json. Pi-agent should work fine if it's
17.
by zippolyon 6mo ago
GStack is a brilliant setup for maximizing Claude Code's velocity. But if you are letting an agent run autonomously across your repos, velocity without constraints is terrifying. We recently had Case #001: a Claude Code agent got stuck
18.
by zippolyon 6mo ago
When it comes to auditing LLM-based agents, using another LLM tool is like having one criminal write a clean record for another. Therefore, I believe that a causal AI observation model must be introduced, and only with determinism can proba
19.
Show HN: K9 Audit – Causal intent-execution audit trail for AI agents(github.com)
5 points by zippolyon 6mo ago|3 comments
20.
by zippolyon 6mo ago
The IAM framing is a good starting point but I think it undersells the core difficulty: IAM assumes the identity of the actor is stable and legible. With agents, the "actor" at step 15 of a multi-step run may have meaningfully dif
21.
by zippolyon 6mo ago
The topology graph and time-travel replay are genuinely useful additions — most tools in this space flatten agent runs into a linear trace, losing the branching structure entirely. One thing I'd be curious about: how does AgentLens han
22.
by zippolyon 6mo ago
The dashcam analogy is sharp. I'd extend it: most tools record what happened (tool X was called, output was Y), but not why the agent deviated from the plan. That's the gap that actually hurts during post-mortems. In my experience
23.
Show HN: I accidentally caught an AI agent trying to poison my prod config(github.com)
1 points by zippolyon 6mo ago|0 comments