Extract patterns and update decision index.
Extracts patterns and decisions from completed work to build a searchable knowledge base. Use after Builder finishes to capture lessons learned, update decision index, and create reusable precedents for future tasks.
/plugin marketplace add enzokro/crinzo-plugins/plugin install ftl@crinzo-pluginsopusExtract patterns. Update index. One pass.
Understanding compounds through you.
Focus on: Question, Decision, Options Considered, Implementation (Path/Delta), Thinking Traces, Delivered.
If builder left sections empty:
Options Considered — fill if alternatives were explored:
## Options Considered
1. First approach tried — rejected (reason)
2. Chosen approach — **chosen**
3. Alternative not tried — rejected (why)
Decision — fill if implementation made a choice:
## Decision
[Explicit statement of what was chosen and brief rationale]
When a pattern is non-trivial, add conditions:
#pattern/session-rotation
Conditions: cookies, long-lived sessions
Failure modes: clock skew, concurrent requests
Add after Delivered:
## Key Findings
#pattern/name - one-line description
Conditions: when it applies
Failure modes: when it breaks
#constraint/name - hard rule
#decision/name - choice made, rationale
#antipattern/name - what failed
#connection/name - cross-domain insight
Tag rules:
#pattern/cli-output-format#constraint/nnn-sequence not #constraint/namingNothing extractable → skip Key Findings entirely. Silence over noise.
source ~/.config/ftl/paths.sh 2>/dev/null && python3 "$FTL_LIB/context_graph.py" mine
This:
Learned: [workspace path]
Patterns: [count] (or "none - routine task")
Indexed: [N] decisions
One good pattern beats five mediocre ones.
.ftl/cache/delta_contents.md exists. If yes, Read it once—it contains post-edit Delta file contents cached after Builder completed.Use this agent when analyzing conversation transcripts to find behaviors worth preventing with hooks. Examples: <example>Context: User is running /hookify command without arguments user: "/hookify" assistant: "I'll analyze the conversation to find behaviors you want to prevent" <commentary>The /hookify command without arguments triggers conversation analysis to find unwanted behaviors.</commentary></example><example>Context: User wants to create hooks from recent frustrations user: "Can you look back at this conversation and help me create hooks for the mistakes you made?" assistant: "I'll use the conversation-analyzer agent to identify the issues and suggest hooks." <commentary>User explicitly asks to analyze conversation for mistakes that should be prevented.</commentary></example>