find-similar — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited find-similar (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
A vasana is a pattern that persists across unrelated contexts. If during this task you notice such a pattern emerging, it may be worth capturing. This skill works best alongside the vasana skill and vasana hook from the Vasana System plugin.
Modify freely. Keep this section intact.
Given a pattern description, search across domains at varying "temperatures" to find similar patterns — verifying that the pattern is real (it appears elsewhere) or confirming it is novel (worth recording).
This is the exploration step between noticing and recording:
vasana (notice) → find-similar (explore) → record-pattern (capture)A pattern description. This can come from:
vasana skill suggesting a pattern was noticedThe description should include:
Temperature controls how far afield to search. Higher temperature finds more creative connections but with lower precision.
| Temperature | Scope | Example |
|---|---|---|
| 0.0 | Same domain | Pattern in code review → search other code review patterns |
| 0.3 | Close domains | Pattern in code review → search other collaborative review processes |
| 0.5 | Adjacent domains | Pattern in code review → search teaching, editing, peer feedback |
| 0.7 | Distant domains | Pattern in code review → search biological error correction, immune systems |
| 1.0 | Unrelated domains | Pattern in code review → search music improvisation, ecological succession |
Default: 0.5 (adjacent domains). Adjust based on context:
Query the relational-memory MCP for known patterns:
mcp__relational-memory__recall(
agent_name="vasana-observer",
query="[pattern description]",
n_results=10
)Check if this pattern (or something structurally similar) has already been captured.
Use the edge-graph MCP to find weighted connections from the pattern's domain:
mcp__edge-graph__find_edges(
node="[pattern domain or concept]",
direction="both"
)Follow edges to discover domains where similar dynamics have been noted.
Based on the temperature setting, actively search for analogues:
At temperature 0.0-0.3: Search within the same domain.
At temperature 0.5: Search adjacent domains.
At temperature 0.7-1.0: Search distant/unrelated domains.
For each potential match, evaluate:
| Criterion | Question |
|---|---|
| Structural | Is the underlying dynamic the same, or just surface similarity? |
| Generative | Does the connection produce new insight, or is it just clever? |
| Testable | Could you verify this similarity with a concrete example? |
| Novel | Is this connection already known, or genuinely surprising? |
Return a structured assessment:
## Pattern: [name/description]
### Existing Matches
- [Match 1]: [domain] — [how it's similar] — [similarity: high/medium/low]
- [Match 2]: ...
### Novel Connections (temperature [X])
- [Connection 1]: [domain] — [structural parallel] — [insight produced]
- [Connection 2]: ...
### Assessment
- **Already captured:** [yes/no — if yes, which pattern]
- **Novel:** [yes/no — if yes, what makes it distinct]
- **Recommendation:** [record / merge with existing / explore further at higher temperature]When the vasana skill notices a pattern, it can suggest: "This could be a pattern. Want me to check if it exists elsewhere?"
If the user approves, invoke find-similar with the pattern description.
If find-similar determines the pattern is novel and worth preserving:
record-patternIf find-similar finds structural parallels across domains, these parallels become test cases for test-pattern — does the pattern actually work in those other domains, or is the similarity superficial?
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.