levi-qiao/longgraph-skill / skills/loop-graph
loop-graph
FlaggedUse only when the user explicitly invokes /loop-graph or names this skill. Compiles a custom loop-graph run. Do not use unless named. Existing runtime nodes are self-contained.
templates/findings.md
View on GitHub<!--
loop-graph template: findings.md — the SCOUT'S SINGLE-WRITER STATE EDGE.
Seeded empty at generation time into the run directory. The scout node writes here;
the executor reads it ON-REFERENCE ONLY (when it hits a ledger pointer), never
every round. The supervisor may read it during audit to verify the executor consumed
findings correctly.
Discipline: State edge, single-writer (scout), read-on-reference (executor).
POINTER CONVENTION — the ledger carries a one-line pointer at the decision point:
blocked-on: findings#<brief-id>
The executor opens this file (or findings/<brief-id>.md) only when it reaches that
row. Most rounds it never reads findings.
CONSUME-AND-RETIRE — once the executor folds the decision into the plan:
1. Record the decision + rationale in the ledger's round log.
2. Move the consumed finding to archive/findings-<brief-id>.md (or delete it).
3. Remove the blocked-on pointer from the ledger.
The findings edge stays O(active briefs), not O(total briefs ever dispatched).
PARALLEL SCOUTS — when multiple scouts run concurrently, each writes its own file
under findings/<brief-id>.md instead of this root findings.md. Each file is
single-writer (its scout). The pointer convention stays the same:
blocked-on: findings/<brief-id>#<section>
On a single-scout run, this root file suffices. Create the findings/ directory only
when concurrency is real.
-->
# {{PROJECT}} — Findings (scout → executor, read-on-reference)
<!-- Findings appear below as the scout completes research briefs. The executor
reads a finding only when it reaches the corresponding blocked-on pointer in
the ledger. Once consumed, the finding is archived (moved to archive/) and the
pointer is removed. -->
(no findings yet)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |