shim-cli launches privacy layer for coding agents

// 2h agoOPENSOURCE RELEASE

shim-cli launches privacy layer for coding agents

shim-cli is an Apache-2.0 local hook and proxy that masks secrets and personal data before coding agents send tool results to models. It supports Claude Code, Codex, Copilot CLI, and VS Code, while reporting session tokens, costs, and detected exposures.

// ANALYSIS

shim-cli tackles a real blind spot in agentic coding: developers often do not know how much context their tools transmit. Its local-first design is compelling, but coverage gaps and best-effort detection mean it complements rather than replaces robust data-loss prevention.

–Claude Code gets the strongest protection, including masking eligible tool results before model access.
–Codex, Copilot CLI, and VS Code currently offer reporting or prompt controls with materially less tool-result coverage.
–Enforce mode can block sensitive prompts, but the default behavior reports some exposures after the fact.
–No account, telemetry, or shim network call makes the tool attractive for privacy-conscious developers.
–The open-source detector and performance tests improve trust, though custom workloads still need independent validation.

// TAGS

shim-clisecuritycoding-agentai-codingcliopen-sourceobservability

DISCOVERED

2h ago

2026-10-18

PUBLISHED

8h ago

2026-10-18

RELEVANCE

9/ 10

AUTHOR

[REDACTED]

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

MODEL1h ago

AgentDecision-4B Makes Agent Loops 89% Cheaper

AgentDecision-4B is a LoRA-tuned Qwen3-4B adapter that handles bounded tool, argument, and completion choices inside agent harnesses, deferring uncertain states to larger language models. The paper reports 81.1–89.0% lower costs and up to 2.5× faster execution on familiar tasks, plus 91.5% success on shifted tasks. [Paper](https://arxiv.org/abs/2610.21412)

RESEARCH1h ago

JevAgentHarness routes agent decisions with confidence gates

JevAgentHarness is an open-source proxy that routes routine agent decisions to fast models while escalating uncertain steps to stronger LLMs. Its AgentDecisionBench research reports major cost and latency reductions using per-choice confidence thresholds and the open AgentDecision-4B model.

OPEN SOURCE1h ago

git-ai brings server-centric Git to agents

git-ai is an MIT-licensed, open-source Git platform built for concurrent coding agents. It centralizes history, sparse data access, landing, and configurable conflict resolution on a server while retaining Git underneath. ([GitHub repository](https://github.com/feder-cr/git-ai))