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 stefanoskarakasis/Product-Marketing-Skills --skill meta-synthesisgit clone --depth 1 https://github.com/stefanoskarakasis/Product-Marketing-SkillsWrote 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/stefanoskarakasis/product-marketing-skills/meta-synthesis)<a href="https://agentmods.dev/skills/stefanoskarakasis/product-marketing-skills/meta-synthesis"><img src="https://agentmods.dev/badge/skills/stefanoskarakasis/product-marketing-skills/meta-synthesis/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/stefanoskarakasis/product-marketing-skills/meta-synthesis"><img src="https://agentmods.dev/badge/skills/stefanoskarakasis/product-marketing-skills/meta-synthesis.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 | $0.00096 | $0.02327 |
| Opus 5.5 | $0.00038 | $0.00931 |
| Sonnet 5.5 | $0.00019 | $0.00465 |
| Haiku 4.5 | $0.00010 | $0.00233 |
Grade A, and why
meta-synthesis 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 7d 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 — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
meta-synthesis — Skill
How This Works
Reads the session log every execution skill already writes to, finds patterns that repeat across 2 or more sessions, and turns confirmed patterns into either a guardrail (for /context/meta-patterns.md) or a brain update (for /foundation/brain.md). This is what makes the rest of the stack compound instead of starting fresh every session.
The skill runs in 5 steps:
Step 0 — Load /context/skill-sessions.md and /context/meta-patterns.md. Block if the session log doesn't exist or has zero rows.
Step 1 — Scan sessions for repeated patterns, both within one skill and across skills.
Step 2 — Rank patterns by confidence (occurrence count) and propose next action for each.
Step 3 — Surface proposed guardrails and brain updates, gated on user approval.
Step 4 — Write approved guardrails to /context/meta-patterns.md and approved brain updates to /foundation/brain.md. Log the session.
Trigger
- When: Detecting patterns across 2+ prior sessions logged by other skills, and turning confirmed patterns into guardrails or brain updates.
- Not for: Real-time feedback during a session — execution skills already check
/context/meta-patterns.mdat their own Step 0. Auditing a single skill's output quality → usemeta-review. One-off pattern lookups — this skill's value is in running it repeatedly. - Example prompts:
- "run meta-synthesis"
- "what patterns are emerging"
- "detect cross-skill signals"
- "update guardrails"
- "compound our learnings"
- "what should we remember"
Inputs
- Args: Timeframe to analyze — all sessions, or a specific window (e.g. last 30 days). Asked in Step 0 if not specified.
- Defaults: All sessions in
/context/skill-sessions.mdif no timeframe is given. - Context keys:
/context/skill-sessions.md— required. The session log every execution skill writes to./context/meta-patterns.md— optional; loaded if it exists, treated as empty otherwise./foundation/brain.md— optional; Sections 1-6 loaded silently as baseline context for proposed updates.- Brain contract: Reads Sections 1-6. Writes: Section named in an approved brain-update proposal only, after explicit user approval (Step 3).
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago Changed · +4 lines · -10 tokens per session 930fcac8c1e0
- 1mo ago Changed · +4 lines 601f9318d40e
- 1mo ago First seen · 211 lines · 106 tokens per session scan A 2e8a1b7ff1df
meta-synthesis is a skill published in the GitHub repository stefanoskarakasis/Product-Marketing-Skills (8 stars, last pushed yesterday), licensed MIT. It adds 96 tokens to every session and 2,327 once invoked, about $0.0004 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
reflect
Per-project self-improvement - reads the .harness ledger and feedback memories, then proposes gated rule/threshold/ADR changes so the project stops repeating mistakes. Run periodically.
remember
Review auto-memory entries and propose promotions to CLAUDE.md, CLAUDE.local.md, or shared memory. Also detects outdated, conflicting, and duplicate entries across memory layers.
okhp3-thread-context-extraction
Extract pasted or uploaded AI chat threads into standalone, actionable Markdown. Use when the user wants to evacuate context from ChatGPT, Claude, Gemini, Perplexity, Microsoft Copilot, Grok, Mistral Vibe, an unknown AI platform, or a mixed transcript; preserve goals, decisions, reasoning, next actions, prompts, code…
report
A way to export a chat card or reply as a standalone HTML report, meaning a file that can open in a browser without the chat application. It preserves the card template, including right-to-left Persian or left-to-right English layouts.
k8s-rightsize
Right-size Kubernetes CPU and memory requests from Prometheus usage (p95 CPU, p98 memory over 7 days), size the node count, tune KEDA autoscaling. Use when asked to rightsize workloads, cut node cost, or "how many nodes do we need".
owasp-security
Deep OWASP reference: Top 10 2025, ASVS 5.0, secure patterns, per-language quirks, LLM Top 10 and agentic AI security. Use for an in-depth security review, auth work, or hardening an LLM or agent app. For the everyday baseline use secure-coding.