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.
npx agentmods add skills/laserphaser/claude-ltm/ltm-savenpx skills add LaserPhaser/claude-ltm --skill ltm-savegit clone --depth 1 https://github.com/LaserPhaser/claude-ltmWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/laserphaser/claude-ltm/ltm-save)<a href="https://agentmods.dev/skills/laserphaser/claude-ltm/ltm-save"><img src="https://agentmods.dev/badge/skills/laserphaser/claude-ltm/ltm-save.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00015 | $0.00415 |
| Opus 5 | $0.00008 | $0.00208 |
| Sonnet 5 | $0.00003 | $0.00083 |
| Haiku 4.5 | $0.00002 | $0.00042 |
Grade A, and why
ltm:save 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 5d 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.
What it actually says
Save a Decision to Memory
The user wants to explicitly record a specific decision. They may provide
the decision as an argument (e.g., /ltm:save chose Stripe for payments)
or describe it in their message.
What to do
-
Read
.memory/DECISIONS.mdto check for existing related entries. -
Parse the user's input to understand the decision being recorded.
-
Check for duplicates — if a similar decision already exists, update it rather than creating a new entry.
-
Add the decision to the appropriate section in
.memory/DECISIONS.md:- Use a
### headingwith today's date - Include up to ~10 lines of context explaining the decision and rationale
- Place it under the most fitting
## Sectionheading (create a new section if none fits — sections are free-form) - If the decision supersedes a prior one, update the old entry in-place and add a Changelog entry
- Use a
-
If the decision is complex and the user provided extensive rationale:
- Create a detail file at
.memory/details/YYYY-MM-DD-<topic>.md - Keep the main file entry concise with a link to the detail file
- Create a detail file at
-
Update the
> Auto-maintained by claude-ltm. Last updated:date. -
Confirm to the user what was saved and where.
Detail file format
When creating a detail file in .memory/details/, use this structure:
---
decision: <title>
date: YYYY-MM-DD
status: ACTIVE
---
<!-- Valid status values: ACTIVE, SUPERSEDED -->
## Context
[Why this decision was needed — 2-3 sentences]
## Decision
[What was chosen and why — concise]
## Alternatives Considered
- Option B: [why rejected]
## Impact
- [Key consequence 1]
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.
- 5d ago First seen · 62 lines · 15 tokens per session scan A 53591e102e23
ltm:save is a skill published in the GitHub repository LaserPhaser/claude-ltm (4 stars, last pushed 4mo ago), licensed MIT. It adds 15 tokens to every session and 415 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.
Other skills, from other repositories
memory-map
Two-tier project memory system. Maintains a global Claude memory snapshot of all ongoing projects AND a per-project STATUS.md file in each project folder that captures last-session state, next actions, blockers, and key locations — so context survives long absences. Operates in two modes. WRITE mode triggers when the…
memory-prune
Clean stale, outdated, or redundant entries from Claude's stored memory. Use this skill when the user says things like "clean up memory", "prune memory", "memory is getting cluttered", "remove old entries", "consolidate memory", or "trim the memory map". Also trigger proactively when memory approaches its limits — 30…
memory-recall
Load and summarise stored project memory at the start of a session, or whenever the user asks what Claude remembers about their work. Use this skill when the user says things like "what do you remember", "what's the context", "where did we leave off", "catch me up", "what's the current state", "show me the memory…
gather-context
Before working on a task, pull the relevant slices of the project's ground truth from the manifest, write a short context brief, and confirm it with the human in one batch. Re-run at each phase boundary, not just at the start.
mem0-dream
Consolidates stored memories by merging duplicates, resolving contradictions, and pruning stale entries. Use when memory count is high, search results feel noisy or repetitive, or periodic cleanup is needed to maintain memory quality.
mem0-vercel-ai-sdk
Mem0 provider for Vercel AI SDK (@mem0/vercel-ai-provider). TRIGGER when: user mentions "vercel ai sdk", "@mem0/vercel-ai-provider", "createMem0", "retrieveMemories", "addMemories", "getMemories", "searchMemories", "mem0 vercel", "AI SDK provider", "AI SDK memory", or is using generateText/streamText with mem0. Also…