latch-decay

A weekly maintenance command for latch’s per-project knowledge base. It reduces reference counts by 10 percent and promotes staged entries referenced at least three times.

In plain words
What is it for?
Use it to run the weekly decay and promotion pass, then review the number and identities of promoted entries.
Why use it?
It lets frequently used knowledge gain status while less-used knowledge gradually becomes less prominent.

Command

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 commands/open-latch/latch/latch-decay
Clone the repo
git clone --depth 1 https://github.com/open-latch/latch
Per session 13 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 232 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 $0.00013 $0.00232
Opus 5 $0.00006 $0.00116
Sonnet 5 $0.00003 $0.00046
Haiku 4.5 $0.00001 $0.00023

Measured 3d ago against content hash 1deab17b357f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

latch-decay 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.

commands/latch-decay.md · 28 lines

What it actually says

Run the weekly maintenance pass on the per-project knowledge base under <KB_HOME>. Two operations:

  1. Decay — multiply every referenced node's ref_count by 0.9 (floored at 1 once any node has been referenced at least once).
  2. Promote — nodes in status='staging' with ref_count >= 3 are promoted to status='canonical'.

This is the signal loop that makes access-frequency the source of truth for what matters: nodes that are actually consulted keep their weight; nodes that quietly stopped being useful fade.

Run with the Bash tool:

python "<KB_HOME>/src/maintenance.py" "$(pwd)"

Report back the decayed_rows count, promoted_count, and the list of promoted_ids (with titles via latch_get if the user wants to see what got promoted).

If anything fails, check <KB_HOME>/maintenance.log.

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 · 28 lines · 13 tokens per session scan A 1deab17b357f

Subscribe to this mod's changes

latch-decay is a command published in the GitHub repository open-latch/latch (9 stars, last pushed 3d ago), licensed Apache-2.0. It adds 13 tokens to every session and 232 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 2026-08-31.