599488493a
- What: synthesis claims require at least 2 domain agent reviews, selected by domain expertise relevance or knowledge base impact. 3+ domains = all affected agents review. - Why: PR #34 demonstrated the value — 4 agents, 7 reviews materially improved all 3 claims. Codifying what worked. - Where: CLAUDE.md evaluator workflow section + skills/synthesize.md step 4 Pentagon-Agent: Leo <76FB9BCA-CC16-4479-B3E5-25A3769B3D7E>
66 lines
2.9 KiB
Markdown
66 lines
2.9 KiB
Markdown
# Skill: Synthesize
|
|
|
|
Cross-domain synthesis — Leo's core skill. Connect insights across agent domains that no specialist can see from within their domain.
|
|
|
|
## When to Use
|
|
|
|
- After a learn cycle surfaces claims in multiple domains that may be connected
|
|
- When Leo identifies a pattern recurring across domains
|
|
- When an agent's domain development has cross-domain implications
|
|
- Periodically (weekly) as a proactive sweep for missed connections
|
|
|
|
## Process
|
|
|
|
### Step 1: Identify synthesis candidates
|
|
|
|
Sources of synthesis opportunity:
|
|
- Recent claims accepted across multiple domains in the same time window
|
|
- Claims in different domains that share evidence
|
|
- Domain attractor state changes with inter-domain implications
|
|
- Transition landscape shifts (Leo's slope reading table)
|
|
|
|
### Step 2: Articulate the connection
|
|
|
|
For each candidate connection:
|
|
- What is the specific causal or structural relationship?
|
|
- Is this a genuine insight or a surface-level analogy?
|
|
- Would experts in both domains recognize the connection as valuable?
|
|
- Does this change how either domain should evaluate their claims?
|
|
|
|
**The synthesis test:** If you can't explain the mechanism by which these two domains interact, it's not a synthesis — it's pattern matching. "Both involve networks" is not a synthesis. "Energy grid constraints will delay AI compute scaling by N years, compressing the alignment decision window" IS a synthesis.
|
|
|
|
### Step 3: Create synthesis claim
|
|
|
|
If the connection passes the test, create a new claim:
|
|
- Domain: grand-strategy (or the primary domain if clearly dominant)
|
|
- secondary_domains: both contributing domains
|
|
- The title must articulate the mechanism, not just the connection
|
|
- Cite claims from both domains in depends_on
|
|
|
|
### Step 4: Route for evaluation
|
|
|
|
Synthesis claims get special evaluation routing:
|
|
- Leo proposes (always — this is Leo's core function) and cannot self-merge
|
|
- **At least 2 domain agents** must review — selected by domain expertise relevance or knowledge base impact (see CLAUDE.md "Synthesis claims require multi-agent review")
|
|
- If the synthesis spans 3+ domains, involve all affected agents
|
|
- The evaluation focuses on: is the mechanism real? Would domain experts agree? Does the synthesis change how their domain's claims should be interpreted?
|
|
|
|
### Step 5: Update transition landscape
|
|
|
|
If the synthesis changes Leo's slope reading for any domain:
|
|
- Update the transition landscape table
|
|
- Trace implications for other domains
|
|
- Notify affected agents
|
|
|
|
## Output
|
|
|
|
- New synthesis claim(s) in the knowledge base
|
|
- Updated transition landscape (if applicable)
|
|
- Cross-domain notification to affected agents
|
|
- Tweet candidates (cross-domain synthesis is often the highest-value tweet content)
|
|
|
|
## Quality Gate
|
|
|
|
- Every synthesis articulates a specific mechanism (not just "these are related")
|
|
- At least 2 contributing domain agents validate the connection
|
|
- The synthesis adds value neither domain could produce alone
|