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 skills add hifriendbot/cogmemai-mcp --skill session-startgit clone --depth 1 https://github.com/hifriendbot/cogmemai-mcpWrote 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/hifriendbot/cogmemai-mcp/session-start)<a href="https://agentmods.dev/skills/hifriendbot/cogmemai-mcp/session-start"><img src="https://agentmods.dev/badge/skills/hifriendbot/cogmemai-mcp/session-start/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/hifriendbot/cogmemai-mcp/session-start"><img src="https://agentmods.dev/badge/skills/hifriendbot/cogmemai-mcp/session-start.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00075 | $0.01015 |
| Opus 5.5 | $0.00030 | $0.00406 |
| Sonnet 5.5 | $0.00015 | $0.00203 |
| Haiku 4.5 | $0.00007 | $0.00102 |
Grade A, and why
session-start 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Session Start with CogmemAi
Run this skill at the start of every session, before responding to the user's first message. It loads the most important memories from prior sessions for the current project, so the assistant can pick up exactly where the last session left off.
When to Use
- The very first turn of a new session, before any other action
- After any context compaction event (you'll see a summary of the prior conversation)
- When the user greets the assistant ("hi", "hey", "morning", "let's continue")
- When switching from one project to another within the same session
- When the user says "what were we working on?", "where did we leave off?", "remind me"
How to Run
Call get_project_context. Optional but useful: pass a context parameter describing the current task, which biases the returned memories toward what's most relevant.
get_project_context({
context: "implementing the new payments flow"
})
The response includes:
- Top memories ranked by importance and recency, scoped to the current project plus any global memories that apply
- A
health_score(0-100) indicating overall memory quality - Any pending tasks from prior sessions
- Active reminders set by the previous session via
set_reminder
What to Do With the Result
-
Read every returned memory carefully. They contain architecture decisions, preferences, gotchas, file paths, and bug fixes that took effort to discover. Treat them as authoritative for what was true when they were written.
-
Verify before acting on memory-derived recommendations. A memory that names a specific file, function, or flag is a claim about the past. If the user is about to act on your recommendation, confirm the artifact still exists. Things get renamed, removed, or never merged.
-
Check pending tasks. If
get_tasks(or the embedded task list) shows in-progress or todo items, surface them to the user as "here's what was open" so the user can choose what to tackle. -
Honor active reminders. If a reminder fires (e.g., "remember to delete the test data on the server"), surface it to the user immediately.
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.
- 1mo ago First seen · 96 lines · 75 tokens per session scan A fd110d993049
session-start is a skill published in the GitHub repository hifriendbot/cogmemai-mcp (7 stars, last pushed today), licensed MIT. It adds 75 tokens to every session and 1,015 once invoked, about $0.0003 per session on Opus 5.5. 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
yantrikdb
Persistent cognitive memory via YantrikDB's MCP server. Semantic recall, remember, belief revision, knowledge graph, contradiction detection, and procedural learning that survive across sessions and agents. Tools are auto-discovered from the server at runtime.
persistent-memory
Give this agent durable memory across sessions using YantrikDB. Use at the start of every session to load prior context, whenever the user or a job references past decisions, people, preferences, or earlier runs, and whenever something worth keeping is learned (a decision, a fact, a working method, a qualified lead, a…
cortadel
Cortadel is a self-hosted, bi-temporal graph memory server for AI agents — one .NET container exposing a REST API and an MCP endpoint over FalkorDB or Memgraph, with intent-aware writes (store/invalidate/delete-entity/touch/resolve), LLM-verified dedup with a negation guard, and hybrid BM25+vector+RRF retrieval…
forget
Retract or permanently erase a Cortadel memory. Use when the user says "forget that...", "X is no longer true", "that's outdated", "delete this memory", or "permanently remove ". Default path is a natural-language retraction that keeps history queryable; hard, unrecoverable deletion only happens on an explicit…
history
Reconstruct what Cortadel memory said as of a past date, or trace a fact's supersession chain (what replaced or invalidated it, and when). Use when the user asks what they believed on a given date, wants to see superseded/historical versions of a fact, or asks how something changed over time. mem0 has no equivalent …
recall
Search Cortadel memory deliberately — broader, filtered, or budgeted retrieval beyond what the automatic per-prompt recall gives. Use when the user asks to look something up in memory, search past context, or wants results scoped by project/tag/type — e.g. "search my memories for X", "what do we know about Y", "recall…