Menu
Unified agent memory. Use /remember to store facts, experience, observations, and mental-models in .agents/memory/.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Use when the user invokes /act on a PR/MR or /act <context> with context ∈ {pr, plan, backlog, harvest}. Resolves threads in product code (or posts a substantive in-thread reply), commits, then closes threads. Never resolve-only. Harvest (collecting threads) lives in /harvest; triage (priority / grouping / wontfix) lives in /backlog. /act is the fix loop, not the collect or triage.
Domain-agnostic AI-native scenario runner (e2e-agent CLI). Scenarios are markdown prompts; evidence is written by the CLI. Use when the user invokes /e2e or asks to run a live scenario test and collect evidence under .e2e/results/. Agents use only e2e-agent commands — never import internal framework modules.
Use when the user invokes /harvest on a PR or runs `harvest:pr` / `harvest:batch`. Collects unresolved PR review threads into .agents/review-debt/harvests/*.jsonl on PR merge (or via workflow_dispatch). Does not run during /act and does not open a harvest bot PR — new files are append-only paths pushed straight to main. Downstream consumers: /backlog (triage), /act (batch fix). Harvest is a one-way collect, not a fix loop.
Actionable improvement items derived from experience and retrospectives. Items live in .agents/backlog/.
Reflect on current session and record experience. /retrospect = remember. --plan = + backlog item. --fix = + implement now.
OpenADT product layer — fetch, proxy, transport choice (SDK default), MCP stub. Use when implementing or debugging ADT requests.
| name | memory-bank |
| description | Unified agent memory. Use /remember to store facts, experience, observations, and mental-models in .agents/memory/. |
Unified agent memory inspired by Hindsight.
All entries are markdown files in .agents/memory/.
| Type | Dir | What | Example |
|---|---|---|---|
| fact | facts/ | Observable truth about the project | "tsdown bundles MCP launcher to dist/" |
| experience | experience/ | Something that happened — mistake, debugging session, decision | "git merge bloated PR #56 history" |
| observation | observations/ | Consolidated pattern from multiple facts/experiences | "Codacy 'N issues (0 annotations)' always means run linter locally" |
| mental-model | mental-models/ | Curated principle or heuristic | "Always rebase, never merge feature branches" |
mental-models → observations → facts + experience
(principles) (patterns) (raw ground truth)
When reasoning, check mental-models first (curated), then observations (consolidated), then raw facts/experience (evidence).
/remember <type> — <content>
Write to .agents/memory/<type>/YYYY-MM-DD-<slug>.md:
---
date: YYYY-MM-DD
tags: [tag1, tag2]
---
<content>
Frontmatter is minimal: only date and tags. Add context: <URL> when there's a relevant link. Omit fields you don't know.
Before starting work, scan .agents/memory/ for relevant entries. Read mental-models first, then observations, then facts/experience as needed.
After a session with multiple related experiences or facts, consolidate them into an observation or mental-model. Delete or archive the raw entries if the consolidation captures everything.
--destination values, or /e2e … <sid> examples. Describe roles: "partial SID", "operator ~/.adtls store".ABC, BHF→ABC) and leave it reads like history — that invents false sessions.SID_CLIENT_USER_LANG and one fixture example (ABC_200_USER_EN) for syntax — not live-run narratives../tmp/ — never system /tmp (see AGENTS.md rule 5)