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/govanai/mem-graph/mem-graph-practicenpx skills add GoVanAI/mem-graph --skill mem-graph-practicegit clone --depth 1 https://github.com/GoVanAI/mem-graphWrote 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/govanai/mem-graph/mem-graph-practice)<a href="https://agentmods.dev/skills/govanai/mem-graph/mem-graph-practice"><img src="https://agentmods.dev/badge/skills/govanai/mem-graph/mem-graph-practice.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 | $0.00080 | $0.00866 |
| Opus 5 | $0.00040 | $0.00433 |
| Sonnet 5 | $0.00016 | $0.00173 |
| Haiku 4.5 | $0.00008 | $0.00087 |
Grade A, and why
mem-graph-practice 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.
Mem-Graph Practice
Use the repository's vendor-neutral contract at
../../../cognitive-os/agent-practice/MEM_GRAPH_AGENT_PRACTICE.md. Announce
that this skill is active and state the resolved project scope.
Bootstrap
- Inspect
git statusbefore editing and preserve unrelated work. - Resolve the exact
project_id. Usecognitive-osfor Cognitive OS governance, experiments, policies, events, roadmap, and program state. Do not infer applicability from cross-project availability. - Prefer one
cognitive_agent_bootstrapcall with a narrow task query,include_global=false, and the exact project. For non-trivial Cognitive OS work, supplycanonical_idsonly when the operator or project has configured them; otherwise omit the field and discover governing candidates. Requestinclude_canonical_content=truewhen configured IDs are used. - Directly verify candidate tracker, scope-boundary, or role records before use. Never copy memory IDs from bundled examples.
- If the bootstrap tool is unavailable, resolve deployment-local canonical
IDs from operator/project configuration or exact-project governing guidance,
fetch each resolved record, read roadmap and active artifacts referenced by
a verified tracker, run
cognitive_policy_lookupforrequest_type/current_canonical_guidance, then run exact-projectcognitive_current_guidance_searchorcognitive_current_guidance_diagnose. - Fetch every selected governing record directly when the bootstrap snapshot did not include its full content or its authority still needs verification.
Use narrow search terms. FTS5 uses AND semantics, so broad compound queries can miss the intended record.
Decide and Act
- Only governing-lane records may influence current-guidance selection. Contextual/ineligible records may explain history or contamination but may not drive the decision.
- Verify canonical role, adoption, scope, applicability, and current evidence. Eligibility and rank do not grant authority; candidate policies stay advisory. System and operator instructions outrank stored guidance.
- Act only within the task's existing authority. Preserve unrelated work and do not commit, publish, release, or perform an unrequested destructive operation.
- Keep global inclusion off unless the task explicitly needs it; report when it is enabled.
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.
- 3d ago First seen · 84 lines · 80 tokens per session scan A 9bed4f82749b
mem-graph-practice is a skill published in the GitHub repository GoVanAI/mem-graph (0 stars, last pushed today), licensed MIT. It adds 80 tokens to every session and 866 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 2026-08-31.
Other skills, from other repositories
dismech-compliance
Skill for analyzing and improving compliance in the dismech knowledge base. Use this skill when checking disorder file completeness, identifying missing fields (ontology terms, evidence, descriptions), understanding weighted priority scoring, and systematically improving knowledge base coverage.
memory-manager
Manage HOT/WARM/COLD memory transitions, Wiki Phase 1/2/3 entity sedimentation, GDPR Art 17 purge. Triggered by hooks/stopfinalize.py + /forget-project command.
hermes
Run the HealthClaw guardrail stack inside Hermes, Nous Research's self-improving AI agent. Same conversational gateway and same compliance goals as the OpenClaw integration, but with Hermes' learning loop — skills that improve from experience, conversation memory across sessions, and native MCP support over Streamable…
cross-examination-planner
Plan ethical, issue-led cross-examination grounded in testimony, documents, contradictions, perception, memory, bias, and admissible impeachment material. Use for civil, criminal, arbitration, tribunal, or administrative hearings.
aget-check-kb
Validate Knowledge Base health by checking structure, frontmatter compliance, and content staleness. Returns conformance score and health status (OK/WARN/CRITICAL).
lians
Use Lians financial-grade agent memory - store and recall facts with a bitemporal model so stale revisions never contaminate context. Use whenever an agent needs persistent memory in finance, healthcare, or legal work, or when a task asks "what did we know on/before ", needs an audit trail, or must erase a data…