From 998a5793b7c09e1e1cada4134fe9fe076e826eb2 Mon Sep 17 00:00:00 2001
From: KT <677465+kevintseng@users.noreply.github.com>
Date: Sun, 16 Aug 2026 21:40:34 +0800
Subject: [PATCH] docs(agents): what an agent must never invent
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Two negative instructions, prompted by studying ctx's agent skill (their
anti-hallucination clauses are the best part of that doc): when memesh is
unavailable or recall returns nothing, say so and work without it — never
fabricate a memory or cite a [mem:id] handle that was not actually
returned; and recall windows are bounded, so a hit count is not a
graph-wide count and an empty result is not proof of absence. A wrong
'remembered' fact arrives wearing the authority of the graph, which makes
it worse than no memory at all.
---
 AGENTS.md                 | 12 ++++++++++++
 dist/skills-manifest.json |  4 ++--
 skills/memesh/SKILL.md    |  7 +++++++
 3 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/AGENTS.md b/AGENTS.md
index fa05c25e..ceb6ebc6 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -52,6 +52,18 @@ host is recallable from all of them. Not installed yet? Follow
   `observation` parameter removes just that fact and keeps the entity active.
   Prefer it over archiving the whole entity.
 
+## What you must never invent
+
+- If MeMesh is unavailable or a recall returns nothing, **say so and work
+  without it** — never fabricate a memory, and never cite a `[mem:id]`
+  handle that was not actually shown to you. A wrong "remembered" fact is
+  worse than no memory: it arrives wearing the authority of the graph.
+- Recall windows are bounded (the `limit` parameter, 30 by default), so do
+  not infer graph-wide counts or "there is no memory about X" from the
+  number of hits one query returns. Absence of results is absence of
+  results, not evidence of absence — vary the wording or narrow by tag
+  before concluding anything.
+
 ## What Claude Code already does — do not double-write
 
 Under Claude Code with the MeMesh plugin, hooks capture automatically:
diff --git a/dist/skills-manifest.json b/dist/skills-manifest.json
index e9e76ec8..5c6fde9f 100644
--- a/dist/skills-manifest.json
+++ b/dist/skills-manifest.json
@@ -113,8 +113,8 @@
     },
     {
       "path": "skills/memesh/SKILL.md",
-      "sha256": "ad76b477c74434f533cc786838fdd6645d734623475c917d5a24326b8c5c83e7",
-      "bytes": 9092
+      "sha256": "04dacfaeab1cb0af2ebe041a8825a44d2ffcae4e192a158217eabdc585721b2e",
+      "bytes": 9498
     }
   ]
 }
diff --git a/skills/memesh/SKILL.md b/skills/memesh/SKILL.md
index 1ba5a6b0..35f390fd 100644
--- a/skills/memesh/SKILL.md
+++ b/skills/memesh/SKILL.md
@@ -56,6 +56,13 @@ clear it.
 Run `memesh briefing` (or `--project <name>`) and answer from it. For specific
 follow-up questions, use `recall`.
 
+**MEMESH UNAVAILABLE or RECALL EMPTY → say so, never invent.** Report that
+memory is unavailable (or found nothing) and continue without it. Never
+fabricate a memory or cite a `[mem:id]` that was not actually returned.
+Recall is bounded by `limit` — a small hit count is not a graph-wide count,
+and an empty result is not proof nothing was stored: vary the wording or
+narrow by tag before concluding.
+
 ## What's Already Automatic (Claude Code Plugin Hooks)
 
 If MeMesh is installed as a Claude Code plugin, these happen **without any action from you**: