try-works/role-model / .agents/skills/recursive-mode
recursive-mode
UnverifiedRepository workflow orchestration skill for staged implementation, locked artifacts, late-phase receipts, and durable memory maintenance. Use when executing recursive-mode runs, resuming a run, locking a phase, or verifying locks.
.recursive/memory/MEMORY.md
View on GitHub# MEMORY.md
<!-- RECURSIVE-MODE-MEMORY:START -->
## Memory Router
This file is the durable memory router for the repository.
It is not a knowledge dump. Store durable memory in sharded docs under `domains/`, `patterns/`, `incidents/`, `episodes/`, `training/`, `skills/`, or `archive/`.
Control-plane docs are not memory docs:
- `/.recursive/RECURSIVE.md`
- `/.recursive/STATE.md`
- `/.recursive/DECISIONS.md`
- `/.codex/AGENTS.md`
- `/AGENTS.md`
- `/.agent/PLANS.md`
## Retrieval Rules
- Read this file before loading any other memory docs.
- Load only the memory docs relevant to the current task.
- If the task may benefit from prior recursive-mode experiential learnings, use this index to identify the relevant docs under `/.recursive/memory/training/` and `/.recursive/memory/domains/`.
- The optional `recursive-training-sync.py` helper is read-only; it prints startup guidance about what to read, but does not modify `MEMORY.md` or the memory plane.
- If the task plans delegated review, subagent help, review bundles, smoke-harness portability work, or capability-sensitive execution, read `/.recursive/memory/skills/SKILLS.md` and then load only the relevant skill-memory shards that are actually present.
- If Phase 8 will need to promote durable lessons, first capture run-local skill usage in the run artifact and only then promote generalized conclusions into skill-memory shards.
- Prefer `Status: CURRENT` docs for planning and execution.
- `Status: SUSPECT` docs may be used as leads, but revalidate them before trust.
- Exclude `STALE` and `DEPRECATED` docs from default retrieval unless doing historical analysis.
## Registry
- `domains/` - stable functional-area knowledge with `Owns-Paths`
- `patterns/` - reusable playbooks and solution patterns
- `incidents/` - recurring failure signatures and fixes
- `episodes/` - distilled lessons from specific runs
- `training/` - extracted experiential learnings promoted from completed recursive-mode runs
- `skills/` - durable skill and capability memory, routed via `skills/SKILLS.md`
- `archive/` - historical or deprecated memory docs
## Freshness Rules
- Durable memory docs must declare the metadata defined in `references/artifact-template.md`.
- Any doc whose `Owns-Paths` or `Watch-Paths` overlaps final changed code paths must be reviewed in Phase 8.
- Affected `CURRENT` docs should be downgraded to `SUSPECT` until revalidated against final code, `STATE.md`, and `DECISIONS.md`.
- If changed paths have no owning domain doc, create one or record the uncovered-path follow-up in `08-memory-impact.md`.
- Training memory docs should keep their canonical content under `/.recursive/memory/training/`, use the memory index as the discovery surface, and record source runs plus watch-path or applicability guidance.
- Skill-memory docs should record source runs, last validated date, environment notes, and current trust/fit guidance.
- If a run materially teaches the repo something about skill availability, delegated-review quality, review-bundle usage, or toolchain fallback behavior, Phase 8 must either create/refresh a skill-memory shard or record why no durable lesson was promoted.
- If the repo itself is a reusable skill/workflow distribution, durable memory must remain generalized. Do not store current-session run residue or temp-environment observations as if they were universal truth.
<!-- RECURSIVE-MODE-MEMORY:END -->
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: 469k | 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: 321k | 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: 321k | 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: 321k | Updated | Claude CodeCodex | Verified | Install |