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/skyfox-io/memex/updatenpx skills add Skyfox-io/Memex --skill updategit clone --depth 1 https://github.com/Skyfox-io/MemexWhat 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 | $0.00079 | $0.00816 |
| Opus 5 | $0.00039 | $0.00408 |
| Sonnet 5 | $0.00016 | $0.00163 |
| Haiku 4.5 | $0.00008 | $0.00082 |
Grade A, and why
update scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 84 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Memex - Update
Wikilink rule: When referencing files in any markdown you write, use [[filename]] format.
Mid-session sync. Flushes status and decisions only. No session-log entry. No hub updates. No _CLOSETS.md refresh. No wikilink checks. For a full close, use /memex:session-end.
Step 1: Resolve paths
Run WORKSPACE_ROOT=$(pwd) && echo "$WORKSPACE_ROOT" via Bash.
Resolve paths via: Config table (if present) > convention (memory/) > search.
If _MANIFEST.md does not exist, tell the user "Memex is not initialized. Run /memex:init first." and stop.
Step 2: Update status.md
- Update "Last updated" to today
- Update "What's In Progress" to reflect current state
- Update "What's Blocked". Resolve unblocked items, add new blockers
- Update "Next Priorities" if they've shifted
Step 3: Log decisions
If meaningful decisions have been made this session, append to decisions.md:
**YYYY-MM-DD** - [decision stated as a fact, one sentence]
Skip if no decisions were made.
Step 4: Annotate superseded decisions
Scan decisions.md for entries where a later entry explicitly overrides an earlier one. Look for language in newer entries referencing older decisions: supersedes, replaces, dropped, no longer, instead of, reverses.
For each match:
- Find the older entry being referenced.
- If not already struck through, wrap it in
~~strikethrough~~. - Append
(superseded YYYY-MM-DD)with the date of the newer entry.
Only annotate when the override is explicit in the text. Do not infer contradictions from topical similarity. That's /memex:lint's job.
Skip this step if decisions.md has fewer than 5 entries.
Step 5: Confirm
Memory updated.
status.md - updated [date]
Decisions: [count logged, or "none"] ([N] superseded entries annotated)
Session still open. Run /memex:session-end to close fully.
Gotchas
- Update is not session-end. It deliberately skips session-log, hub summaries,
_CLOSETS.md, wikilink check, typed-edge graph, and lock clearing. If the session is actually ending, use/memex:session-end. Otherwise the next session-start will see a stale state and warn. - Decisions.md is append-only here. Compression lives in
/memex:consolidate. If the file is approaching its 100-line cap, surface a one-line nudge in the confirmation; don't compress in-line. - Supersession detection is text-explicit only (Step 4). Do not infer from topical similarity. If two decisions about the same project don't reference each other, they're both live until
/memex:lintflags the contradiction. - No idempotency check here (unlike session-end Step 3). Running
/memex:updatetwice in a row will refresh the "Last updated" date both times even if nothing else changed. That's fine for a checkpoint primitive. But means update is not a substitute for session-end's idempotent close.
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.
- 3d ago First seen · 84 lines · 79 tokens per session scan A 338d753f7ca8
update is a skill published in the GitHub repository Skyfox-io/Memex (17 stars, last pushed 1mo ago), licensed MIT. It adds 79 tokens to every session and 816 once invoked, about $0.0004 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.
Other skills, from other repositories
obsidian
Comprehensive guidelines for Obsidian.md plugin development including ESLint rules from eslint-plugin-obsidianmd v0.4.1, TypeScript best practices, memory management, API usage (requestUrl vs fetch), UI/UX standards, popout window compatibility, community.obsidian.md submission process, and Scorecard optimization. Use…
paid-ads
When the user wants help with paid advertising campaigns on Google Ads, Meta (Facebook/Instagram), LinkedIn, or other ad platforms. Also trigger on "PPC", "paid media", "ROAS", "CPA", "ad campaign", "retargeting", "audience targeting", "Google Ads setup", "Facebook Ads strategy", "LinkedIn Ads". Covers campaign…
sales-enablement
When the user wants to create sales collateral, pitch decks, one-pagers, objection handling docs, or demo scripts. Also trigger on "sales deck", "pitch deck", "one-pager", "leave-behind", "objection handling", "ROI calculator", "demo script", "talk track", "sales playbook", "proposal template". For competitor battle…
aia-generation
Run an AI impact assessment — structured intake, risk analysis, UK regulatory classification (UK GDPR / DPA 2018, ICO AI guidance, sector-specific UK regulation), EU AI Act analysis where EU nexus exists, policy consistency diff, and recommendation with conditions. Uses the house-style structure learned from the seed…
cold-start-interview
Run the cold-start interview — learns your UK AI governance practice and writes /.claude/plugins/config/uk-legal-plugins/ai-governance-legal-uk/CLAUDE.md from your AI policy, a reference impact assessment, and key vendor AI agreements. Use when the practice profile is missing or contains [PLACEHOLDER] markers, or when…
integration-management
Post-completion M&A integration tracker — phased workplan, consent tracking, contract assignment at scale, weekly status reports. Initialises from whatever deal artefacts are available (purchase agreement, deal summary, closing checklist) and connects to deal-context.md and closing-checklist.yaml from the M&A…