memory-denied-write

A memory configuration for an agent that has a write tool listed but is explicitly forbidden to use it.

In plain words
What is it for?
It is for testing that denied write access prevents memory changes and avoids creating project memory files.
Why use it?
It ensures the denylist is treated as part of write-capability checks, so the agent receives read-only memory instead of misleading write instructions.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/tintinweb/pi-subagents/memory-denied-write
Clone the repo
git clone --depth 1 https://github.com/tintinweb/pi-subagents
Per session 17 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 304 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00017 $0.00304
Opus 5 $0.00009 $0.00152
Sonnet 5 $0.00003 $0.00061
Haiku 4.5 $0.00002 $0.00030

Measured 3d ago against content hash b23a0fb2c3b7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

memory-denied-write scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

test/fixtures/.pi/agents/memory-denied-write.md · 26 lines

What it actually says

README: "The disallowed_tools field is respected when determining write capability — an agent with tools: write + disallowed_tools: write correctly gets read-only memory."

This is the combination neither memory-readonly.md (no write tool at all) nor memory-readwrite.md (write tool, nothing denied) exercises: the write tool IS in the tools: set, so a naive capability check sees it and hands the agent the read-write memory prompt — which also appends write/edit to the tool names, widening the set the agent asked for. The denylist still filters the registry, so the visible symptom is an agent instructed to write memory files with no tool to do it; the invisible one is the tool-name widening.

Scope project is safe here precisely because the read-only branch creates no memory directory — if this fixture ever flips to the read-write branch, it would also start writing into the repo, which is the loudest possible signal.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 26 lines · 17 tokens per session scan A b23a0fb2c3b7

Subscribe to this mod's changes

memory-denied-write is an agent published in the GitHub repository tintinweb/pi-subagents (1,047 stars, last pushed 5d ago), licensed MIT. It adds 17 tokens to every session and 304 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.